OUTLINE

1
Association Studies of Asthma
Edwin K. Silverman, M.D., Ph.D. Channing
Laboratory and Pulmonary and Critical Care
Division Brigham and Womens Hospital
2
Association Studies of Asthma Potential and
Pitfalls

• Promising Potential
• Asthma susceptibility genes IL4RA (Ober 2000)
• Asthma severity genes B2AR (Weir 1998)
• Asthma pharmacogenetic genes ALOX5 (Drazen 1999)
• Disappointing Reality
• No proven asthma genetic determinants
• Inconsistent results of association studies

3
Association of the ADAM33 Gene with Asthma and
Bronchial Hyperresponsiveness(Van Eerdewegh et
al., Nature 2002 418426)

• Positional Cloning Approach
• Genome Scan Linkage with STR Markers
• Association Studies of Comprehensive Set of
  Positional Candidate Gene SNPs
• Study Population
• 460 Families with an Asthmatic Sib Pair
• All Caucasian 362 from UK, 98 from US
• Phenotypes
• Asthma Physician Diagnosis and Current Med Use
• Asthma BHR Includes PC20 lt 16 mg/ml

4
Linkage Analysis Results Van Eerdewegh et al.
(2002)

• Two peaks with Max LOD 2.94 to Asthma Dx on chrom
  20p
• Asthma Dx plus BHR Max LOD 3.93 on chrom 20p
• 1 LOD support interval of 4.3 cM (2.5 Mb) 40
  genes identified in 1 LOD support interval

5
SNP Haplotype Association Analysis Results Van
Eerdewegh et al. (2002)

• Imputed haplotypes of pairs of adjacent SNPs in
  ADAM33
• Case-Control Analysis More significant results
  noted p values as low as 0.00004 in Combined
  Sample and 0.000003 in UK Sample
• TDT Analysis More impressive results than
  single SNP analysis 8 haplotypes significant at
  p lt 0.005

6
Association of the ADAM33 Gene with Asthma Key
Issues(Van Eerdewegh et al., Nature 2002
418426)

• Important Lessons
• Apparent success of positional cloning in a
  complex disease
• Utility of haplotype analysis in susceptibility
  gene identification
• Uncertainties
• Replication Will ADAM33 be replicated in other
  populations?
• Function What is the key functional variant or
  set of variants in ADAM33?

7
B2AR and Asthma Genetics

• Multiple polymorphisms in the Beta-2 Adrenergic
  Receptor have been identified, including
  nonsynonymous SNPs at amino acid 16 (nucleotide
  46) and amino acid 27 (nucleotide 79)
• B2AR is located on chromosome 5q
• Case-control association studies have
  demonstrated inconsistent relationships between
  B2AR and asthma-related phenotypes including IgE,
  Airway hyperresponsiveness, FEV1, Steroid
  dependence, Nocturnal peak flow, Bronchodilator
  responsiveness, and Qualitative assessment of
  asthma severity.

8
CAMP Genetics Ancillary Study Population

• Childhood Asthma Management Program Participants
  
• 1041 children with mild to moderate asthma
• Extensive baseline and longitudinal phenotypic
  information before randomization to drug therapy.
• Recruitment of Parents
• All available parents of CAMP children were
  invited to provide blood samples
• Phenotypic information not collected on parents
• DNA samples from complete parent/child trios were
  available for 651 nuclear families inclusion of
  sib pairs within the CAMP Study provided a total
  of 707 parent/child trios within these 651
  families (Total number of genotyped parents and
  probands 1999).

9
Why Use CAMP Trios Instead of Cases/Controls?

• Extensive phenotype data available on CAMP
  participants, but no control subjects included
  Cost saving of reduced genotyping in case/control
  overwhelmed by cost and impracticality of
  recruiting/phenotyping controls
• Utility of family-based data in assessing
  haplotypes
• Ability to include racially diverse subjects may
  reduce power, but will not cause pop
  stratification

10
Baseline CAMP Phenotypes Analyzed

• Asthma Diagnosis
• Atopy IgE, Eosinophil Count, Allergy Skin Tests
• Pulmonary Function/Severity FEV1 ( Predicted),
  FVC ( Predicted), PEFR (AM and PM), Asthma
  Symptom Score
• Airway Responsiveness FEV1 Increase post-BD
  (Absolute volume, Baseline Predicted), PC20

11
(No Transcript)
12
Drysdale (2000) B2AR SNPs/Haplotypes

• Identified 13 SNPs with 12 haplotypes
• Three common haplotypes in Caucasians (All
  others had frequencies lt 2)
• One other haplotype was common in
  African-Americans (25) and Hispanics (10)

13
Drysdale SNPs in B2AR Analyzed with BEST Program
14
B2AR SNPs
Drysdale
htSNPs
CAMP
-1023


-709
-709
-709
-654
-654
-654
-468
-468

-406
-406
-367
-47
-47
-20
46
46
46
79
79
252
252
252
491
491
491
523
523
523
15
SNPs Included in ?2-AR Haplotyping

• Nucleotide Gene Location
• -709 (C/A) 5
• -654 (G/A) 5
• -47 (T/C) BUP AA19 Cys/Arg
• 46 (G/A) AA16 Gly/Arg
• 79 (C/G) AA27 Gln/Glu
• 252 (G/A) synonymous
• 491 (C/T) AA164 Thr/ Ile
• 523 (C/A) synonymous

16
Genotyping of 8 B2AR SNPs

• Selected from 13 Drysdale SNPs (PNAS 2000) to
  distinguish three common haplotypes in Whites
• SNaPshot genotyping was inconsistent
• Sequenom MassExtend genotyping was performed
• Pedigree inconsistencies were assessed with
  Pedcheck and resolved (Removed 0-12 peds/marker).
• Hardy-Weinberg equilibrium was assessed

17
Allele Frequencies and Hardy-Weinberg Equilibrium
of B2AR SNPs in Parents of CAMP Children
SNP
H-W in All Parents
H-W in White Parents
Number of Parents Typed
Rare Allele Frequency
-709
n.s.
n.s.
1263 (97.8)
0.002
-654
P 0.03
n.s.
1256 (97.2)
0.36
-47
n.s.
n.s.
1241 (96.1)
0.37
46
n.s.
P 0.047
1249 (96.7)
0.40
79
P 0.04
n.s.
1237 (95.7)
0.37
252
n.s.
n.s.
1171 (90.6)
0.23
491
n.s.
n.s.
1259 (97.4)
0.01
523
n.s.
n.s.
1275 (98.7)
0.20
18
(No Transcript)
19
Family-Based Association Analysis TDT vs.
FBAT
Transmission/Disequilibrium Test (Spielman, et
al., 1993) -Assesses transmission from
heterozygous parents to affected
offspring -Both parents are required one
affected per pedigree in standard test of
association (Modified to include missing
parents and haplotypes in TRANSMIT by Clayton,
1999) -McNemars Test compares transmitted vs.
nontransmitted alleles -Used for Qualitative
CAMP Phenotypes Family-Based Association Test
(FBAT) (Horvath and Laird, 2000) -Able to
perform analysis with incomplete family
trios -Used for Quantitative CAMP Phenotypes
20
CAMP B2AR Association Analysis Analytical
Approach

• Population All CAMP families with confirmatory
  analysis in Whites only
• Qualitative Phenotypes
• Single SNPs and Haplotypes (3 most common)
• TRANSMIT
• Quantitative Phenotypes
• Single SNPs and Imputed Haplotypes
• FBAT

21
Family-Based Association Analysis with B2AR
SNPs Qualitative Asthma-Related Phenotypes
Qualitative Phenotype
SNP -654
SNP -47
SNP 46
SNP 79
SNP 252
SNP 491
SNP 523
FEV1 lt 80
P lt 0.001
P 0.001
P lt 0.001
P 0.002
N.S.
N.S.
N.S.
FVC lt 80
P 0.02
N.S.
P 0.007
N.S.
N.S.
N.S.
N.S.
Sx Score gt 1
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
PC20 lt 4
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
BD gt 200 ml
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
BDBase gt 15
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
P 0.015
BDPred gt 15
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
P 0.005
Pos Skin Test
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
Eos gt 500
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
IgE gt 100
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
Asthma Dx
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
22
TRANSMIT Single SNP Analysis in CAMP Issues

• Small numbers of pedigrees for FEV1 lt 80
  phenotype (n 20)
• Lack of robustness to population stratification
  Only for incomplete parental data

23
Family-Based Association Analysis with B2AR
SNPs Quantitative Asthma-Related Phenotypes
Quantitative Phenotype
SNP -654
SNP -47
SNP 46
SNP 79
SNP 252
SNP 491
SNP 523
FEV1 ( Pred)
P 0.012
N.S.
P 0.015
N.S.
N.S.
N.S.
N.S.
FVC ( Pred)
N.S.
N.S.
P 0.09
N.S.
N.S.
N.S.
P 0.09
Sx Score
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
Ln PC20
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
BD Abs Volume
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
P 0.07
BD Baseline
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
P 0.012
BD Predicted
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
P 0.008
Pos Skin Tests
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
Total Eos
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
Log(Total IgE)
N.S.
P 0.08
N.S.
N.S.
N.S.
N.S.
N.S.
AM PF Mean
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
PM PF Mean
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
24
Key Quantitative Phenotype Values in CAMP
Asthmatic Children by B2AR Genotype
Quantitative Phenotype
SNP 46 Arg/Arg
SNP 46 Arg/Gly
SNP 46 Gly/Gly
FEV1 ( Predicted)
103.2 11.8
103.0 12.8
102.6 12.8
FVC ( Predicted)
105.7 11.3
107.2 13.0
105.6 13.2
Quantitative Phenotype
SNP 523 C/C
SNP 523 C/A
SNP 523 A/A
BD (Abs Volume)
0.17 0.13
0.15 0.15
0.12 0.09
BD ( Baseline)
11.3 9.7
8.8 9.1
7.5 5.8
BD ( Predicted)
9.7 7.1
7.7 7.2
6.9 5.3
25
Bronchodilator Responsiveness in CAMP Children
Stratified by 523 SNP Genotype
26
Haplotypes in Gene Localization

• Definition The specific alleles at loci
  inherited from one parent typically refers to
  closely grouped loci on one chromosome.
• Why study haplotypes?
• More informative than individual SNPs
• Reflect evolutionary history/linkage
  disequilibrium pattern more accurately
• May allow identification of key combinations of
  SNPs

27
Comparison of Common B2AR Haplotypes in CAMP and
Drysdale et al.
Haplotype Frequencies

• Haplotypes CAMP Drysdale Drysdale
• Caucasian
  African-Americans
• CGCGGGCC 36.3 48.3 6.3
• CATACGCC 35.4 33.0 29.7
• CGTGCACA 18.2 13.2 31.3

28
Family-Based Association Analysis of B2AR
Haplotypes in CAMP

• Qualitative Phenotypes TRANSMIT
• Robust variance estimation
• Assess 3 most common haplotypes in global
  statistic (gt 10 frequency)
• Quantitative Phenotypes FBAT
• Impute haplotypes with PHASE in parents and
  probands separately
• Identify pedigree inconsistencies/removed 52
  pedigrees
• FBAT on imputed haplotypes (as alleles)

29
Haplotype Imputation Using Phase PrettyBase
Files
30
Haplotype Imputation Using PhasePhase Output in
CAMP Parents
31
(No Transcript)
32
Family-Based Association Analysis of B2AR
Haplotypes in CAMP Issues

• Qualitative Phenotypes (TRANSMIT)
• Impact of small number of trios for key
  phenotypes
• 20 families with transmissions to affected
  offspring for FEV1 lt 80 pred.
• Chi square vs. Bootstrap p values
• Impact of rare SNPs (e.g. -709, 491)
• Results not changed significantly
• Similar results in Caucasians only

33
Family-Based Association Analysis of B2AR
Haplotypes in CAMP Issues

• Quantitative Phenotypes (FBAT)
• Accuracy of haplotype imputation ??
• Consistency of haplotype imputation
• Only two inconsistent subjects in 3 PHASE runs
• Impact of removing 8 of families
• 32/52 removed families had pedigree errors for
  single SNPs
• 52/651 (8) inconsistent pedigrees for all
  subjects 40/477 (8) in whites only
• TDT can have increased false-positive rate when
  inconsistent pedigrees are removed (Gordon AJHG
  2001)
• Similar results in Caucasians only
• Development of haploFBAT

34
Multiple Comparisons in CAMP B2AR Association
Analysis

• 8 SNPs
• Two very rare SNPs
• Significant LD how many independent
  observations?
• 12 Quantitative and 11 Qualitative Phenotypes
• Some obviously correlated (3 BD measures)
• Is this 23 phenotypes, 12 phenotypes, or a
  smaller number?

35
Pearson Correlation Coefficients Between CAMP
Proband Phenotypes
Quantitative Phenotype
FEV1
FVC
Sx Score
PC20
BDABS
BDBASE
BDPRED
NPOS
EOS
IGE
AMPF
PMPF
FEV1 ( Pred)
--



FVC ( Pred)
0.82
--
Sx Score
-0.02
0.06
--
Log PC20
0.10
-0.06
-0.18
--




N.S.
N.S.
BD Abs Volume
-0.01
0.19
0.16
-0.34
--
BD Baseline
-0.06
0.15
0.17
-0.34
0.88
--
BD Predicted
0.11
0.28
0.16
-0.33
0.89
0.97
--
Pos Skin Tests
0.01
0.02
0.12
-0.23
0.15
0.09
0.10
--
Total Eos
0.00
0.04
0.08
-0.26
0.20
0.25
0.24
0.25
--
Log(Total IgE)
-0.02
0.05
0.09
-0.36
0.26
0.22
0.22
0.44
0.49
--
AM PF Mean
-0.05
-0.08
-0.04
0.13
0.17
-0.09
-0.11
0.15
0.02
0.13
--
PM PF Mean
-0.02
-0.03
-0.02
0.14
0.21
-0.09
-0.10
0.13
0.01
0.11
0.98
--
36
Interpretation of CAMP Phenotype Correlation
Coefficients

• Many correlations are significant, but only a few
  are very high
• FEV1 and FVC (0.82)
• AM and PM Peak Flow (0.98)
• BDBASE, BDPRED, and BDABS (0.88-0.97)
• Biologically reasonable correlations are seen
• Atopy Eosinophils, IgE, and Positive Skin
  Tests
• BD Response, Atopy, and PC20
• How many independent phenotypes are there?
• Development of new approaches to minimize
  multiple comparisons (e.g. grouping phenotypes)
• Importance of replication

37
Family-Based Association Analysis with B2AR
Discussion

• Significant evidence for association of several
  single B2AR SNPs and B2AR haplotypes with FEV1 lt
  80 predicted, but only 20 affected individuals
  below this threshold raises concerns about
  robustness of this finding.
• Some evidence for association of B2AR -654 and
  46 SNPs with FEV1 as quantitative phenotype.
  Effect size based on mean values by genotype is
  modest.
• Significant evidence for association of 523 SNP
  and B2AR haplotypes with bronchodilator
  responsiveness as a qualitative and quantitative
  phenotype.
• With multiple phenotypes studied, importance of
  FEV1 findings is uncertain, but different
  variants in B2AR could exert minor influences on
  asthma severity and bronchodilator
  responsiveness.

38
COLLABORATORS

• CAMP Genetics Study Scott Weiss, Jeffrey
  Drazen, Nan Laird, David Kwiatkowski, Jody Senter
  Sylvia, Ross Lazarus, Christoph Lange, Dawn
  DeMeo, Helen Lyon, Kelan Tantisira, Steve Lake,
  Benjamin Raby