genes, physical activity and obesity actions and … physical activity and obesity actions and...
TRANSCRIPT
Genes, physical activity and obesity actions and interactions
Ruth Loos
Professor and Program Director
Genetics of Obesity and Related Metabolic Traits Program The Charles Bronfman Institute of Personalized Medicine
The Mindich Child Health and Development Institute Department of Preventive Medicine
Icahn School of Medicine at Mount Sinai New York
[email protected] ISBNPA - 2013 Annual Meeting, Ghent, Belgium, 22-25 May, 2013
Overview
• Where do genes fit in ?
• Genetics epidemiology of physical activity
• Heritability
• Candidate gene approach
• Genome-wide scans
• Gene-physical activity interaction
• Summary & conclusions
Overview
• Where do genes fit in ?
• Gene1cs epidemiology of physical ac1vity
• Heritability
• Candidate gene approach
• Genome-‐wide scans
• Gene-‐physical ac1vity interac1on
• Summary & conclusions
Is physical activity heritable ?
Heritability studies
Twin studies Iden,cal / MZ
Fraternal / DZ
Family studies
Mul,-‐genera,onal
Is physical activity heritable ?
Heritability studies
Twin studies Family studies
h2 = 0 -‐ 78% h2 = 0 – 57%
Reasons for the wide range -‐ Difference in assessing physical ac1vity -‐ Age of study popula1on -‐ Methods for es1ma1ng heritability -‐ Sample size, and thus sta1s1cal power for analyses
The classical twin study design to assess heritability
Basic sta,s,c Intra-‐pair correla1ons
1.6 1.6
1.7
1.8
1.9
2
1.7 1.8 1.9 2 Twin 1 – height (m)
Twin 2 – height (m)
r MZ= 1.0
r DZ= 0.50
Basic assump,ons MZ twins: 100% genes in common 100% shared environment in common DZ twins: 50% genes in common 100% shared environment in common
More refined: structural equa1on modeling: V = Va + Vc + Ve
Quick and simple: Falconer’s formula: h2 = 2[rMZ – rDZ]
A large twin study with objectively assessed physical activity
TwinsUK • 419 pairs of MZ twins • 352 pairs of DZ twins Physical ac,vity assessment • 7-‐day monitoring (96% had >4 days) • Heart rate monitors • accelerometer
Marcel den Hoed
PA 82% adhere to the physical ac1vity guidelines
A large twin study with objectively assessed physical activity
Descrip,ves
Residuals PAEE twin 1
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
-‐1.5 -‐1.0 -‐0.5 0.0 0.5 1.0 1.5
Residuals PAEE twin 2
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
-‐1.5 -‐1.0 -‐0.5 0.0 0.5 1.0 1.5
MZICC = 0.47 DZICC = 0.22
A large twin study with objectively assessed physical activity Intra-‐pair correla,ons
Explaine
d varia
nce
0.00
0.20
0.40
0.60
0.80
1.00
1.20
PAEE Sedentary Light MVPA
gene1c factors unique environmental factors
A large twin study with objectively assessed physical activity
Structural equa,on modeling
Heritability of daily physical ac,vity 43-‐49%
Overview
• Where do genes fit in ?
• Gene1cs epidemiology of physical ac1vity
• Heritability
• Candidate gene approach
• Genome-‐wide scans
• Gene-‐physical ac1vity interac1on
• Summary & conclusions
Candidate gene approach
= hypothesis-‐driven, based on the current understanding of the biology and pathophysiology of the trait/disease
Epidemiology Physiology/Biology
Animal studies
Butler et al. Nature Neuroscience 2001
High Fat diet
Kelly et al. J. Neurosci 1998
-/-
-/-
+/-
+/-
+/+
+/+
Candidate gene approach
DRD2 (dopamine receptor D2)
MC4R (melanocor@n 4 receptor)
Reward and aversion Craving and need
Common genetic variation - SNPs
GGC GCC GGA CGC TCC ATC TCC TGG GCC TGG TGG GCA TGT GTG Gly Ala Gly Arg Ser Ile Ser Trp Ala Trp Trp Ala Cys Val
GGC GCC GGA CGC TCC ATC TCC TGG GCC TGG TGG GCA TGT GTG
GGC GCC GGA CGC ACC ATC TCC TGG GCC TGG TGG GCA TGT GTG GGC GCC GGA CGC TCC ATC TCC TGG GCC TGG TGG GCA TGT GTG
GGC GCC GGA CGC TCC ATC TCC TGG GCC TGG TGG GCA TGT GTG GGC GCC GGA CGC TCC ATC TCC TGG GCC TGG TGG GCA TGT GTG
GGC GCC GGA CGC ACC ATC TCC TGG GCC TGG TGG GCA TGT GTG GGC GCC GGA CGC ACC ATC TCC TGG GCC TGG TGG GCA TGT GTG
GGC GCC GGA CGC TCC ATC TCC TGG GCC TGG TGG GCA TGT GTG GGC GCC GGA CGC ACC ATC TCC TGG GCC TGG TGG GCA TGT GTG
A
A A
A
T
T T
T T
T
Gly Ala Gly Arg Thr Ile Ser Trp Ala Trp Trp Ala Cys Val
T/T
A/T
A/A
T/T
A/T
SNP à Single Nucleo,de Polymorphism
Genetic associations
The ‘addi,ve’ model tests whether the A-‐allele increases PAEE significantly.
25.2
26.6
27.9
24
25
26
27
28
29
T/T A/T A/A Genotype
PAEE(kJ/kg/day)
Human studies
Candidate gene approach
MC4R
• 669 individuals of the Québec Family Study • Physical activity assessed by 3-day diary, every 15min
Human studies
Candidate gene approach
MC4R
100
200
300
400 500
600 700
800
Total Moderate-to- strenuous Inactivity
Ph
ysic
al A
ctiv
ity
Cou
nts
C/C C/T T/T p = 0.08
p = 0.005
p = 0.01
Human studies
Candidate gene approach
PCK1
*sex, age, age-‐group, standardised height, maturity and country
• 2,101 boys and girls from European Youth Hearth Study
• 1,255 9-‐y olds
• 846 15-‐y olds
• Denmark and Estonia
• Physical ac,vity -‐ MTI accelerometer
Human studies
Candidate gene approach
PCK1
• 669 individuals of the Québec Family Study • Physical activity assessed by 3-day diary, every 15min
0
540
580
620
660
PCK1 SNPs
Physical Ac,vity cou
nts
0 1 2
p = 0.87 p = 0.57 p = 0.68 p = 0.65 p = 0.67 p = 0.33 p = 0.87
p = 0.95 p = 0.63 p = 0.07
*sex, age, age-‐group, standardised height, maturity and country
Candidate gene approach – literature overview
Reward and aversion
Craving and need
Effect on ability, pain
• Melanocor1n 4 receptor (MC4R) • Lep1n receptor (LEPR)
• Dopamine 2 receptor (DRD2)
Genes, exercise, and physiological factors, Eco J.C. de Geus & Marleen H.M. de Moor, in GeneAc and Molecular Aspects of Sport Performance, Claude Bouchard and Eric P. Hoffman (eds), 2011
Human studies
• Loos et al. (IJO, 2005): Associa1on with moderate-‐to-‐strenuous PA (p=0.005) in the QFS.
• Stefan et al. (IJO, 2002): Associa1on with total PA in Pima Indians (p=0.007) • de Moor et al. (MSSE, 2009) Associa1on with leisure 1me PA in Americans (p=0.0005)
• Simonen et al. (Physiology & Behavior, 2009): Associa1on with physical ac1vity, sports par1cipa1on, and occupa1onal ac1vity (p=0.05) in the QFS and HERITAGE.
• Calcium sensing receptor (CASR)
• Angiotensin-‐conver1ng enzyme (ACE)
• Lorentzon et al. (EJE, 2009): Associa1on with hours spent on physical ac1vity per week in adolescent girls (p=0.01)
• Winnicki et al. (AJMG, 2004): Associa1on with leisure 1me physical ac1vity per week in hypertensives (p=0.001)
Overview
• Where do genes fit in ?
• Gene1cs epidemiology of physical ac1vity
• Heritability
• Candidate gene approach
• Genome-‐wide scans
• Gene-‐physical ac1vity interac1on
• Summary & conclusions
Genome-wide scans
= hypothesis-‐genera1ng, to iden1fy new, unan1cipated, genes and thus expand view of biology and physiology of disease/trait
Epidemiology Physiology/Biology
Genome-‐wide analyses
Genome-‐wide linkage analyses aim to iden@fy a crude chromosomal loca@on of the gene or genes associated with a phenotype of interest
Genome-‐wide associa,on analyses aim to iden@fy an exact chromosomal loca@on of the gene@c variant associated with a phenotype of
interest
Genome-wide scans
Genome-wide linkage studies
• 1,030 sibling of 319 Hispanic families
• Physical ac,vity -‐ accelerometer
Genome-wide association studies
STAGE 1 -‐ Discovery stage
• Determine genotype of up to 2,500,000 gene1c variants (SNPs)
• Case-‐control or cohort studies
• Large amount of tests à significance threshold p < 0.000001
STAGE 2 - Replication/follow-up stage • 1-‐100 SNPs that represent top loci
• Case-‐control or cohort studies – at least equal size of discovery panel
• Meta-‐analysis of stage 1 & 2 results
Published Genome-‐wide Associa,ons
06/2006 06/2007 06/2008 06/2009 06/2010 06/2011 06/2012
NHGRI GWA Catalog www.genome.gov/GWAStudies
>1,700 published GWA at p<5x10-‐8
Genome-wide association for physical activity
Aim: Large-‐scale genome-‐wide meta-‐analysis on five physical ac,vity traits
Outcomes of interest: 1. Moderate-‐to-‐vigorous intensity leisure 1me ac1vi1es 2. Sedentary behavior during commu1ng 3. Sedentary behavior at work 4. TV viewing 5. Other sedentary behavior at home
Genome-‐wide genotypes: 2,5 million gene1c varia1ons across the autosomal genome
Study design Stage 1 (discovery): 33 studies with required data were invited, up to 82,000 individuals Study-‐specific analyses plans sent out (customized by ques1onnaire) Meta-‐analyses of summary sta1s1cs (for each of the 2,5 million variants)
N stage 1 82,231 15,076 61,612 28,695 27,627
Study design Stage 2 (follow-‐up): >40 studies with required data have been invited, expect data from up to 80,000 individuals Follow-‐up of 56 gene1c variants with promising associa1on (p < 10-‐6) in stage 1
Loci with p<10-‐6
6 6 15 17 12
Marcel den Hoed
Genome-wide association for physical activity
Sedentary commu,ng to work (vs ac,ve commu,ng to work)
Genes for physical activity
Candidate gene studies Even though animal studies have iden1fied credible proteins that affect physical ac1vity, gene1c variants in the encoding genes are not (yet) convincingly associated with daily physical ac1vity levels in humans .
Genome-‐wide linkage studies Even though one study provided sugges1ve confirma1on for the MC4R locus, results have not been firmly confirmed.
Genome-‐wide associa,on studies Large-‐scale meta-‐analysis is on-‐going;
-‐ stage 1 results are promising
-‐ stage 2 results will soon be available !
à First results are emerging, but much more work is needed !
Overview
• Where do genes fit in ?
• Gene1cs epidemiology of physical ac1vity
• Heritability
• Candidate gene approach
• Genome-‐wide scans
• Gene-‐physical ac1vity interac1on
• Summary & conclusions
Genes & physical activity
Does physical ac,vity anenuate the associa,on between gene and outcome ? Does genotype anenuate the associa,on between physical ac,vity and outcome ?
Physical activity, FTO and obesity-susceptibility
• 17,619 men and women of the EPIC-‐Norfolk study (40-‐75y), with self-‐reported physical ac1vity levels.
25
25.5
26
26.5
27
27.5
28
All Inac1ve Ac1ve
BMI (kg/m
2 )
CC CT TT CC CT TT CC CT TT
P = 3.1 x 10-‐20
P = 1.5 x 10-‐13
P = 8.5 x 10-‐13
EPIC-‐Norfolk (n=20,374)
Vimaleswaran KS, et al. Am J Clin Nutr 2009;90:425-‐8
Overview of the literature on FTOxPA interaction and BMI
Greek adol. (n=499)
Scon RA, et al. Eur J Hum Genet 2010 Aug 18
Inter 99 (n=5,922)
Interac,on No interac,on
HAPI Heart (n=702)
DPS (n=479)
NHS98 (n=4,210)
Malmö PP (n=15,925)
Botnia PP (n=2,511)
HELENA (n=752)
NFBC86 (n=4,780)
KSNP (n=8,842)
TRAILS (n=1,230)
LACHY & APEX (n=744)
ULSAM (n=1,115)
NFBC66 (n=4,022)
BCAMS (n=3,503)
MDC (n=4,810)
Andreasen CH, et al. Diabetes 2008;57:95-‐101
Rampersaud E, et al. Arch Intern Med 2008;168:1791-‐7
DESIR (n=4,640)
Cauchi S, et al. J Mol Med 2009;87:537-‐46
Cauchi S, et al. J Mol Med 2009;87:537-‐46
Jonsson A, et al. Diabetologia 2009;52:1334-‐8
Jonsson A, et al. Diabetologia 2009;52:1334-‐8
Kaakinen M, et al. Am J Epidemiol 2010;172:653-‐65
Ruiz JR, et al. Arch Pediatr Adolesc Med 2010;164:328-‐33
Tan JT, et al. Diabetes 2008;57:2851-‐7
Sonestedt E, et al. Am J Clin Nutr 2009;90:1418-‐25
Jacobsson JA, et al. BMC Med Genet 2009;10:131
Lee HJ, et al. Clin Chim Acta 2010;11:1716-‐22
Xi B, et al. BMC Med Genet 2010;11:107
Liem ET, et al. Am J Clin Nutr 2010;91:321-‐8
Liu G, et al. BMC Med Genet 2010;11:57
Lappalainen TJ, et al. Obesity 2009;17:832-‐6
Study design
• Contact all cohorts that have published on the FTO locus (and that are known to have physical
ac1vity data).
• ‘Harmonise’ physical ac1vity data across cohorts (i.e. ques1onnaire, diary, accelerometry, …).
à Dichotomise: ac1ve vs sedentary – custom defini1on for each cohort
• Ask cohorts to (re-‐)analyse data according to detailed analysis plan.
Data collec,on & analysis
• 45 studies – 218,166 individuals
• Meta-‐analyses of the interac1on term
• Stra1fied meta-‐analyses of ac1ve and sedentary individuals
Kilpeläinen et al (PLoS Medicine 2011)
De novo meta-analysis of all available data
Tuomas Kilpeläinen
Results
FTO influences BMI in inac1ve people only
Overall interaction term:
b3 = - 0.14 kg/m2 [95%CI -0.23 to -0.04]
P = 0.005
à The BMI-increasing effect of FTO is less pronounced in active individuals than in sedentary individuals; i.e. for each additional risk allele the increase is 0.14kg/m2 less.
0
1000
2000
3000
4000
≤6 7 8 9 10 11 12 13 14 15 16 ≥17 Genetic predisposition score
N
24.5
25.5
26.5
27.5
28.5
BM
I (k
g/
m2)
0.15 kg/m2/allele or 444 g/allele p = 1.54x10-22
Gene1c predisposi1on score on BMI in 20,000 individuals of the EPIC-‐Norfolk study
Li et al AJCN 2010 & PLoS Medicine 2010
Twelve obesity-susceptibility loci – interaction with physical activity
0
1000
2000
3000
4000
≤6 7 8 9 10 11 12 13 14 15 16 ≥17 Genetic predisposition score
N
24.5
25.5
26.5
27.5
28.5
BM
I (k
g/
m2)
0.21 kg/m2/allele or 592 g/allele p = 3.62x10-18
0.13 kg/m2/allele or 378 g/allele p = 7.8x10-21
Active
Inactive
Gene1c predisposi1on score on BMI in 20,000 individuals of the EPIC-‐Norfolk study
Li et al AJCN 2010 & PLoS Medicine 2010
Twelve obesity-susceptibility loci – interaction with physical activity
Overview
• Where do genes fit in ?
• Gene1cs epidemiology of physical ac1vity
• Heritability
• Candidate gene approach
• Genome-‐wide scans
• Gene-‐physical ac1vity interac1on
• Summary & conclusions
Summary
Heritability Based on a large-‐scale twin study with objec1ve PA measurements, 43-‐49% of the inter-‐individual varia1on can be explained by gene1c difference.
Candidate gene approach Animal studies have iden1fied genes that play a role in PA (reward, craving, ability). However, these varia1on in these candidate genes does not (yet) associate with PA in humans. à Study more variants, in larger studies
Genome-‐wide scans On-‐going large scale genome-‐wide associa1on studies have iden1fied loci that show promising associa1on with PA (in par1cular sedentary behavior). These results need to be confirmed.
Gene-‐physical ac@vity interac@on Physical ac1vity atenuates the BMI-‐increasing effects of obesity-‐suscep1bility loci by 30-‐40% à Even those gene1cally predisposed to obesity benefit from physical ac1vity.
Collaborators and acknowledgements
Tuomas Kilpeläinen, PhD Assistant professor Novo Nordisk Founda1on Center for Basic Metabolic Research Copenhagen, Denmark
Marcel den Hoed, PhD Researcher Department of Medical Sciences, Molecular Epidemiology and Science for Life Laboratory Uppsala University Uppsala, Sweden
Marilyn Cornelis, PhD Research Associate in Nutri1on Harvard School of Public Health Boston, MA, USA