Transcript George Davey Smith.ppsx

THE USE OF GENETIC VARIANTS AS TOOLS FOR
EPIDEMIOLOGISTS
George Davey Smith
MRC Integrative Epidemiology Unit
University of Bristol
The principle of Mendelian randomization
(B)
(A)
CRP
SNP
G
E
Y
U
CRP
CHD
U
CHD risk according to duration of current Vitamin E
supplement use compared to no use
RR
2
1.5
1
0.5
0
0-1 year
2-4 years
5-9 years
Rimm et al NEJM 1993; 328: 1450-6
>10 years
Vitamin E supplement use and risk of Coronary Heart Disease
1.1
1.0
0.9
0.7
0.5
0.3
Stampfer 1993
Rimm 1993
RCTs
Stampfer et al NEJM 1993; 328: 144-9; Rimm et al NEJM 1993; 328: 1450-6; Eidelman et al
Arch Intern Med 2004; 164:1552-6
Vitamin E levels and risk factors: Women’s Heart
Health Study
Childhood SES
Manual social class
No car access
State pension only
Smoker
Daily alcohol
Exercise
Low fat diet
Obese
Height
Leg length
Lawlor et al, Lancet 2004
“Well, so much for antioxidants.”
Mendelian randomization
In genetic association studies the laws
of Mendelian genetics imply that
comparison of groups of individuals
defined by genotype should only differ
with respect to the locus under study
(and closely related loci in linkage
disequilibrium with the locus under study)
Genotypes can proxy for some modifiable
risk factors, and there should
be no confounding of genotype by
behavioural, socioeconomic or
physiological factors (excepting those
influenced by alleles at closely proximate
loci or due to population stratification)
Mendel in 1862
Conventional observational epidemiology
Confounders
(Factors associated with both exposure and outcome,
including unmeasured confounders)
and/or
Reverse causation
(Disease alters the modifiable exposure of interest,
rather than vice versa)
Modifiable exposure
(e.g. CRP)
Outcome (e.g. CHD)
Conventional observational epidemiology
Confounders
(Factors associated with both exposure and outcome,
including unmeasured confounders)
and/or
Reverse causation
(Disease alters the modifiable exposure of interest,
rather than vice versa)
Modifiable exposure
(e.g. CRP)
Outcome (e.g. CHD)
It is often impossible to exclude confounding and /or
reverse causation as an explanation for observed
exposure/outcome associations
Mendelian randomization approach
Confounders
(Factors associated with both exposure and outcome,
including unmeasured confounders)
and/or
Reverse causation
(Disease alters the modifiable exposure of interest,
rather than vice versa)
Instrumental
variable
(Genetic variant e.g.
SNP in CRP gene)
Modifiable exposure
(e.g. CRP)
Outcome (e.g. CHD)
Mendelian randomization approach
Confounders
(Factors associated with both exposure and outcome,
including unmeasured confounders)
and/or
Reverse causation
(Disease alters the modifiable exposure of interest,
rather than vice versa)
Instrumental
variable
(Genetic variant e.g.
SNP in CRP gene)
Modifiable exposure
(e.g. CRP)
Outcome (e.g. CHD)
Instrumental variable analysis in MR study
Confounders
(Factors associated with both exposure and outcome,
including unmeasured confounders)
and/or
Reverse causation
(Disease alters the modifiable exposure of interest,
rather than vice versa)
Instrumental
variable
(Genetic variant e.g.
SNP in CRP gene)
Modifiable exposure
(e.g. CRP)
Outcome (e.g. CHD)
Associations of IL6, CRP and fibrinogen (top panel)
and their prediction of CHD (bottom panel)
IL6R Genetics Consortium Emerging Risk Factors Collaboration. Interleukin-6 receptor pathways in coronary heart disease: a
collaborative meta-analysis of 82 studies. Lancet 2012;379:1205–1213
C-Reactive Protein Coronary Heart Disease Genetics Collaboration (CCGC). Association between C
reactive protein and coronary heart disease: mendelian randomisation analysis based on individual
participant data. BMJ 2011;342:d548
CRP CHD Genetics Consortium. Association between C reactive protein and coronary heart
disease: mendelian randomisation analysis. BMJ 2011
FOR FIBRINOGEN see Davey Smith et al Does Elevated Plasma Fibrinogen
Increase the Risk of Coronary Heart Disease?: Evidence from a Meta-Analysis of Genetic
Association Studies. Arterioscler Thromb Vasc Biol 2005; 25: 2228-2233.
Uric acid and IHD using SLC2A9 as an
instrument
Palmer T et al. Association of plasma uric acid with ischaemic heart disease and blood
pressure: mendelian randomisation analysis of two large cohorts BMJ2013;347:f4262
BMI and uric acid using FTO, TMEM
and MC4R as instruments
Palmer T et al. Association of plasma uric acid with ischaemic heart disease and blood
pressure: mendelian randomisation analysis of two large cohorts BMJ2013;347:f4262
Genetic Effects vs.
cross-sectional observation
Limitations
• Reintroduced confounding through pleiotropy
Two categories of pleiotropy
•
•
•
•
Spurious
Relational
Vertical
Type II
•
•
•
•
Genuine
Mosaic
Horizontal
Type I
Gruneberg H. An analysis of the “pleiotropic” effects of a lethal mutation in the rat.
Proc R Soc London, B 1938:125:123-44
Hadorn E. Developmental genetics and lethal factors. Methuen and Company,
London 1961
Wagner GP, Zhang J. The pleiotropic structure of the genotype – phenotype map: the
evolvability of complex organisms. Nature Reviews Genetics 2011; 12: 204-213.
Tyler AL, Asselbergs FW, Williams SM, Moore JH. Shadows of complexity: what
biological networks reveal about epistasis and pleiotropy. Bio Essays 2009; 31: 220227.
Hodgkin J. Seven Types of Pleiotropy. Int. J. Dev. Biol. 1998;42: 501-505.
Approaches from econometrics
• Interact instrument with a second exogenous
variable which modifies (optimally
qualitatively) effect of instrument on
intermediate phenotype
Card D. Using geographic variation in college proximity to estimate the return from
schooling. NBER Working Paper 4483, 1993
GxE in an exposure propensity
example: how does alcohol intake
influence the risk of disease?
Metabolism of alcohol
Ethanol
Acetaldehyde
ADH
Acetic acid
ALDH
CYP2E1
* Mainly occurs in the liver, but some activity is also present in the oral cavity and digestive tract
ALDH2 genotype by alcohol consumption,
g/day: 5 studies, n=6815
60
Alcohol g/day
50
Men
40
Women
30
20
10
0
*1*1
Chen, Davey Smith et al, PLoS Med 2008
*1*2
*2*2
Relationship between characteristics and ALDH2 genotype
Age
70
Smoker
70
60
Percent
60
50
40
50
30
40
20
2*2/2*2
2*2/2*1
1*1/1*1
2*2/2*2
BMI
40
2*2/2*1
1*1/1*1
Cholesterol
250
mg/dl
kg/m2
30
20
200
10
150
0
2*2/2*2
2*2/2*1
1*1/1*1
Takagi et al, Hypertens Res 2002;25:677-81
2*2/2*2
2*2/2*1
1*1/1*1
ALDH2 genotype and systolic blood pressure
Chen et al, PLoS Medicine 2008
Risk of upper aerodigestive cancer by ADH1B genetic variation,
stratified by drinking intensity, rare allele carriers versus common allele
homozygous genotype
Hashibe et al, Nature Genetics 2008
Meta-analysis of association between CHRNA RS1051730 variant and BMI stratified by
smoking status (Freathy et al, 2011).
0.05 kg/m2 [95%CI: -0.05, 0.18]
-0.23kg/m2 [95%CI: -0.13, -0.31]
-0.10 kg/m2 [95%CI: -0.03, -0.18]
P for interaction = 0.0001
Approaches from econometrics
• Interact instrument with a second exogenous
variable which modifies (optimally
qualitatively) effect of instrument on
intermediate phenotype
• Use of multiple instruments
Card D. Using geographic variation in college proximity to estimate the return from
schooling. NBER Working Paper 4483, 1993
Murray MP. Avoiding Invalid Instruments and Coping with Weak Instruments.
Journal of Economic Perspectives 2006;20:111–132
Pleiotropy? Use of multiple instruments in Mendelian
randomization approaches …
Gene 1
Gene 2
Exposure
Outcomes
Confounders; reverse causation; bias
Does body fat increase bone mineral density?
FTO
MC4R
Percent body
fat
BMD
Confounders; reverse causation; bias
Timpson, Davey Smith and Tobias, JBMR 2009
Effect of 9 SNPs from 6 genes on LDL cholesterol and on CHD risk
Ference BA et al. Effect of Long-Term Exposure to Lower Low-Density Lipoprotein Cholesterol
Beginning Early in Life on the Risk of Coronary Heart Disease : A Mendelian Randomization
Analysis. JACC 2012 doi: 10.1016/j.jacc.2012.09.017
SNP associations with uric acid and gout
Yang Q, Kottgen A, Dehghan A, et al. Multiple Genetic Loci Influence Serum Urate Levels and Their
Relationship With Gout and Cardiovascular Disease Risk Factors. Circulation Cardiovascular Genetics
2010;3:523-30
Approaches from econometrics
• Interact instrument with a second exogenous
variable which modifies (optimally
qualitatively) effect of instrument on
intermediate phenotype
• Use of multiple instruments
• Sensitivity analysis using standard methods
(bias is to OLS) and SSIV/JIVE (bias to null)
Card D. Using geographic variation in college proximity to estimate the return from
schooling. NBER Working Paper 4483, 1993
Murray MP. Avoiding Invalid Instruments and Coping with Weak Instruments.
Journal of Economic Perspectives 2006;20:111–132
Approaches from econometrics and
beyond
• Interact instrument with a second exogenous
variable which modifies (optimally
qualitatively) effect of instrument on
intermediate phenotype
• Use of multiple instruments
• Sensitivity analysis using standard methods
(bias is to OLS) and SSIV/JIVE (bias to null)
• Bidirectional instrumentation
Bidirectional or Reciprocal MR
Possible to interrogate pathways
from both directions
BMI (exposure) and CRP (outcome)
Timpson NJ et al. International Journal of Obesity 2011; 35, 300–308.
CRP (exposure) and BMI (outcome)
Timpson NJ et al. International Journal of Obesity 2011; 35, 300–308.
Limitations
• Reintroduced confounding through pleiotropy
• Reintroduced confounding through LD
Limitations
• Reintroduced confounding through pleiotropy
• Reintroduced confounding through LD
• Low statistical power (and weak instrument
bias)
- Multiple instruments (using e.g. CUE, LIML)
- Allele scores
- Two sample MR / IV
Limitations
•
•
•
•
•
•
Reintroduced confounding through pleiotropy
Reintroduced confounding through LD
Low statistical power
Lack of variants to serve as proxy measures
Canalization / developmental compensation
Complexity of associations and failure to
understand what is being instrumented for
Developments of MR
•
•
•
•
Multiphenotype MR
Non-linear associations
“Biomarker demendalization”
Mediation, e.g .epigenomic MR (two step, two
step two sample, etc)
• Anonymous MR
• Data mining with MR instruments
• Hypothesis free causality