IWO bijeenkomst - 17 april - Prof. Dr. A.G. Uitterlinden
1. André G Uitterlinden
Genetic Laboratory
Department of Internal Medicine
Department of Epidemiology
Department of Clinical Chemistry
Human Genomics Facility, HuGe-F
Erasmus MC Genomics Core facility
www.glimdna.org
IWO nascholing, Jaarbeurs, Utrecht, 17 April, 2019
Rol van Genetica bij Osteoporose
Copyright Prof. Dr. A.G. Uitterlinden
2. Companies have sponsored my attendance at some meetings:
- Illumina
- Amgen
- Roche
- MSD
- Calico
My research is funded by public funds from NWO, ZonMW, EU, NIH,
University funds, and charity funds
I receive salary only from Erasmus Medical Centre
I do not hold any stock
Disclosure of Speaker’s Interest
IWO nascholing, Jaarbeurs, Utrecht, 17 April, 2019
Copyright Prof. Dr. A.G. Uitterlinden
3. -Complex Genetics and Genomics
-Genetics of osteoporosis
-Mendelian Randomization studies
-GWAS progress and polygenic risk profiling
-GSA consortium
-Replication and collaboration
-Concluding remarks
IWO nascholing, Jaarbeurs, Utrecht, 17 April, 2019
Rol van Genetica bij Osteoporose
Copyright Prof. Dr. A.G. Uitterlinden
4. The Beauty of Bone…
“The Bone Chair”
-Design: Joris Laarman (NL,1979)
-Just auctioned (6 march 2019) at Christies, London for 825.000 euro
-Created using software to design lighter car-parts in industry
Copyright Prof. Dr. A.G. Uitterlinden
5. We differ from each other…
DNA variation causes differences in:
Development
Appearance
Behaviour
Ageing
Diseases
Copyright Prof. Dr. A.G. Uitterlinden
7. This is what happens
when there are NO POLYMORPHISMS
Why is the study of DNA polymorphisms important ?
Evolution Forensics
Disease
Slide stolen from Prof Axel Themmen
DTC fun
Copyright Prof. Dr. A.G. Uitterlinden
8. ! OPTIMAL EPIDEMIOLOGICAL DESIGN:
A single-centre, longitudinal population-based cohort study of normal elderly Dutch,
started in 1990, with 25 years of follow-up
! LARGE:
Total = 20,000 men and women of age 45 yrs
! VERY DEEP PHENOTYPING:
5 Follow-up measurements with ~1,500 per
subject each time : height, bmi, brain MRI,
DXA, cholesterol level, blood pressure,
glucose, etc. etc. etc.
! ETHNICALLY HOMOGENEOUS:
99% White Caucasian
! EXTENSIVE GENOMICS DATA AVAILABLE:
GWAS, RNA expression (array+ NGS), DNA methylation (450K), Whole Exome
Sequencing, 16S microbiome, telomeres, mitochondrial DNA, metabolomics,..
“ERGO: Erasmus Rotterdam Gezondheid Ouderen”
“The Rotterdam Study”Overview of sample numbers with “omics” datasets across the 3 Rotterdam Study (RS) cohorts with the
number and type of measurement for each omic method
Genomics data type Total Datapoints/sample RS I* RS II* RS III*
Number Type
GWAS SNP data 11,502 40,000,000 SNPs 6291 2157 3054
Exome array 3,183 250,000 SNPs 3183 – –
Whole exome sequencing (WES) 3,778 693,000 Variants 3778 – –
Whole genome sequencing (WGS) 96 3,000,000 Variants 96 – –
Genome wide expression (array) 881 25,000 Genes – – 881
Genome wide expression (RNA Seq) 829 18,000,000 Reads – 500 329
Genome wide DNA methylation 1,600 450,000 CpG’s 100 500 1000
Telomere length (PCR) 1,800 1 – 1800 – –
Mitochondrial DNA (PCR) 500 1 – 500 – –
Microbiome 16S rRNA (faeces) 2,000 500 OTU’s – – 2000
Metabolomics (NMR/UPLC MS) 1,826 4000 Metabolites 1826 – –
Metabolomics (NMR “Nightingale”) 5,381 228 Metabolites 2880 663 1838
Serum protein profile** 9,820 35 Proteins 3812 2542 3466
Total ‘omic’ data points in RS: 43,196 × 62,422,765 = 2,696,413,756,940 (2.7 x 1010 )
SNP single nucleotide polymorphism,
CpG a two-nucleotide position (C next to G on the same strand) of which the C can be methylated;
OTU operational taxonomic unit
*RS1, First cohort of the Rotterdam Study; RS2, Second cohort of the Rotterdam Study; RS3, Third cohort of the Rotterdam Study
**Total estradiol, total testosterone, sex hormone-binding globulin, dehydroepiandrosterone, dehydroepiandrosterone sulfate, androstenedione, 17-
hydroxyprogesterone, cortisol, corticosterone, 11-desoxycortisol, vitamin D, thyroid stimulating hormone, free T4, interleukins, C-reactive protein, Insulin-like growth
factor 1, insulin, iron, ferritin, transferrin, fibrinogen, homocysteine, folic acid, riboflavine, pyridoxine, SAM/SAH ratio, cobalamine, Lp-PLA2, Fas/Fas-L, abeta42/40
(Samples x Datapoints)
Copyright Prof. Dr. A.G. Uitterlinden
9. Human Genomic Life Course Epidemiology
Rotterdam Study
Age (yr)
Bone
Mineral
Density
Bone
growth
Peak BMD Bone Loss
50 7525 100
EPOSCALEUR AGGO
Osteoporosis:
Low BMD, fractures
men
women
DNA/RNA
collections
for OMICS
studies
Maternal genotype
Paternal genotype
Environmental factors
“Chronological” vs. “Biological” Ageing
bone phenoptyes
B-Proof intervention
Birth Death
Bone as an Example...
Off-spring
GenRCopyright Prof. Dr. A.G. Uitterlinden
10. Clinical Expression:
Risk Factors:
Fracture Risk
Bone Strength Impact Force Fall Risk
“OMICS”: DNA RNA, METHYLATION, MICROBIOME, etc
BMD Quality Geometry
Osteoporotic fracture is a “complex” phenotype:
Environmental factors: diet, exercise, ...
Hip fx
Wrist fx
Vertebral fx
etc.
Age, Sex, Age-at-Menopause, Height, OA, etc.
dynamic genomics data: cause/effect?genetic data: cause
Copyright Prof. Dr. A.G. Uitterlinden
11. The next challenge: Environmental Factors
The field needs: standardization, harmonization, replication
HOLLAND BELGIUM
> 1100 mg/day < 500 mg/day
Dietary Calcium intake
Geographical distance: <100km
Foto: Barbara Obermayer-Pietsch Foto: Stuart Ralston
Copyright Prof. Dr. A.G. Uitterlinden
12. EffectSize
Frequency Genetic Variant
rare, monogenic common, complex
Next-Generation Sequencing
(WES/WGS of reference sets)
+
Arrays/Imputation
rare common
smallbig
Genetics: the architecture of diseases/traits :
study designs to identify “risk” alleles
Genome-Wide
Association
Study (GWAS)
few “big” effects of
common alleles
(ApoE, CFH)
Whole Exome
Sequencing (WES)
Copyright Prof. Dr. A.G. Uitterlinden
13. Per 12 May 2018:
• 3,379 publications
• 61,620 unique SNP-trait
associations.
(www.ebi.ac.uk/GWAS )
GWAS ….drinking from the firehose
Copyright Prof. Dr. A.G. Uitterlinden
14. Regulatory
sequences
GWAS identifies mostly (common) regulatory variants
Efforts with a focus on
genes/coding variants:
-WES/WGS
-exome chip
> new arrays with
enhanced clinical content
(e.g., GSA, PMRA)
Gene
Copyright Prof. Dr. A.G. Uitterlinden
15. GEFOS collaboration has generated many GWAS discoveries for BMD…
Slide by Fernando Rivadeneira
Copyright Prof. Dr. A.G. Uitterlinden
16. Largest BMD GWAS over time…
In 2012, GWAS of DXA-BMD in 30,000
individuals from GEFOS, including
CHARGE, identified 56 loci
Estrada et al. Nat Gen 2012
In 2017, GWAS of eBMD in 140,000
individuals from UK Biobank interim
release identified 203 loci (153 novel)
Kemp et al. Nat Gen 2017
5.6% trait variance explained
12% trait variance explained
Slide by John Morris, McGill University, Canada
Copyright Prof. Dr. A.G. Uitterlinden
17. – 1,103 conditionally independent SNPs from 515 loci (301 novel) – 2x the previous study!
– 20% of the trait variance explained – 1.5x increase!
Slide by John Morris; Morris J, Kemp J et al. Nature Genetics (2019)
UK Biobank: Largest BMD* GWAS so far…
in 426,824 White-British participants
*estimated BMD from heel QUS: gSOS
Copyright Prof. Dr. A.G. Uitterlinden
18. *Morris et al., An atlas of genetic influences on osteoporosis in humans and mice. Nature Genetics, 2019
February 2019 GWAS of BMD (as estimated by heel
quantitative ultrasound; =gSOS)
Many genetic effects: >500
*Several low frequency with relatively large effect size and
*Many high frequency with modest effects size
Copyright Prof. Dr. A.G. Uitterlinden
19. Study short name Country Ncases Ncontrols Ntotal
AGES Iceland 1458 1727 3185
AOGC Australia 685 1113 1798
BPROOF Netherlands 715 1483 2198
CHS US 519 2742 3261
DeCODE Iceland 1836 14560 16396
EGCUT-I Estonia 217 4296 4513
EGCUT-II Estonia 71 1717 1788
EPICNOR UK 2937 17726 20663
ERF Netherlands 260 1342 1602
FHS US 1520 2782 4302
GOOD Sweden 273 597 870
HEALTHABC US 308 1353 1661
HKOS Hong Kong 79 627 706
MROS US 918 3555 4473
PROSPER Netherlands 426 4816 5242
RS-I Netherlands 2163 3574 5737
RS-II Netherlands 932 1220 2152
RS III Netherlands 505 2421 2926
SOF US 1611 1698 3309
TUK123 UK 839 4111 4950
UKBB UK 14492 130563 145055
WGHS US 1832 20498 22330
WHICT US 1058 647 1705
WHIOS US 1603 989 2592
YFS Finland 611 975 1586
Total 37857 227116 254973
Discovery: 37,857 cases and
227,116 controls;
Replication: 147,200 fracture
cases and 150,085 controls
(23andMe)
Largest GWAS of (any-type of) fracture to date
comprising 185K cases and 377K controls
Slide by Fernando Rivadeneira (Trajanoska K, et al., BMJ, 2019)
Copyright Prof. Dr. A.G. Uitterlinden
20. Fracture risk GWAS identified 15 loci all of which
are established BMD loci
2p16.2 3p22.1 6q22.33 6q25.1 7q31.31 7q21.3 7p14.1 7p12.1 9q34.11 10q21.1 11q13.2 14q32.12 17q21.31 18p11.21 21q22.2
SPTBN1 CTNNB1 RSPO3 ESR1 WNT16 CPED1 C7orf76 SHFM1 STARD3NL GRB10 COBL FUBP3 MBL2/DKK1 LRP5 RPS6KA5 SOST DUSP3 MEOX1 FAM210A RNMT ETS2
Slide by Fernando Rivadeneira (Trajanoska K, et al., BMJ, 2019)
Copyright Prof. Dr. A.G. Uitterlinden
22. A coordinated roadmap of Integrated functional
assessments will translate into clinical applications
Slide by Fernando Rivadeneira
Copyright Prof. Dr. A.G. Uitterlinden
23. Mendelian Randomization
SNPs as Instrumental Variables:
• principle: alleles segregate and are randomly
inherited from parents to offspring (Mendelian laws)
approach similar to RCTs
• alleles are distributed independent of confounders,
i.e. socio-economic and life-style factors
• inherited genotypes are not changed by a disease (or time)
no reverse causation
• SNPs can explain modest proportion of variance
large sample sizes needed for MR
Copyright Prof. Dr. A.G. Uitterlinden
24. Effect (OR) of Genetic Variants for Risk Factors, on Fracture Risk by MR
Slide by Fernando Rivadeneira (Trajanoska K, et al., BMJ, 2019)
Copyright Prof. Dr. A.G. Uitterlinden
26. Can environmental factors decrease genetic effect?
The case of Age-related Macula Degeneration (AMD)
Risk of Late AMDRisk/protective factor
30% - 3x increasecurrent smoking
30% increasebody mass index
‘mediterranean’ diet 25-40% decrease
physical exercise 40% decrease
Data from E3 and various other consortia 2017 slide provided by Prof. Caroline Klaver
Copyright Prof. Dr. A.G. Uitterlinden
27. Genetic testing predicts life-time risk of Late Age-
related Macula Degeneration (AMD)
Rotterdam Study; Buitendijk et al. 2014; slide provided by Prof. Caroline Klaver
Copyright Prof. Dr. A.G. Uitterlinden
28. Never smoked Smokers (past and current)
62%
17%
8%
7%
91%
30%
11%
10%
Smoking and genetic risk for AMD
Genetic risk in
Non-Smokers
Genetic risk in
Smokers
Rotterdam Study
slide provided by Prof. Caroline Klaver
Copyright Prof. Dr. A.G. Uitterlinden
29. Ho et al. Arch Ophthalmol. 2011; Rotterdam Study I N=8000
Good news: dietary anti-oxidants decrease genetic risk
for AMD…
slide provided by Prof. Caroline Klaver
Copyright Prof. Dr. A.G. Uitterlinden
30. BMD Variance Explained
0%
5%
10%
15%
20%
25%
30%
35%
40%
Age, Sex,
Weight,
Height
All FRAX
Risk Factors
2,094 30,000 150,000 456,000 h2SNP
Sample Size for Genetic Studies
Richards, Lancet
2008
Zheng, Nature
2015
Kemp, Nature
Genetics
2017
Morris,
Nature Genetics
2018
Slide by Brent Richards, McGill University, Monreal, Canada; abstract to ASBMR/ASHG 2018
Note:
-variance explained is done in UKBB. It does not include family history or ≥3 drinks per day.
-DXA BMD in first two papers; eBMD (from heel ultrasound) in last two papers
Estimated total amount
of variance
explained by SNPs
Copyright Prof. Dr. A.G. Uitterlinden
31. UK Biobank
N=502,639
Individuals
Passing
Phenotype and
Genotype QC
N=426,811
UK Biobank
Training Set
N=341,449
UK Biobank
Model Selection
Set
N=5,335
UK Biobank
Test Set
N=4,741
UKB Genotyped
N=488,366
Pass QC
N=486,369
White-British
Ancestry
Subset
N=440,348
SOS Available
& Pass SOS
QC
N=480,521
Phenotype
Quality
Control
Genotype
Quality
Control
Figure 1. Overall Study Design
GWAS and
Training of
PRS Models
Selection of top
PRS Model
Define gSOS
CLSA
N = 6,704
Mr Os USA
N = 4,657
SOF
N = 3,426
PRS: Polygenic Risk Score. QC: Quality Control
Mr Os Sweden
N = 1,880
Test Performance of gSOS in NOGG Screening Program
Slide by Brent Richards, McGill University, Monreal, Canada; unpublished
Copyright Prof. Dr. A.G. Uitterlinden
32. Eligible for NOGG-Based Screening
(>50 years, with at least one risk factor)
CRF Based FRAX to
calculate
10-year probability of major
osteoporotic fracture
Women with prior fragility fracture
CRF Based FRAX: Moderate Risk
CRF Based FRAX: High Risk
BMD Based FRAXBMD Based FRAX: Low Risk BMD Based FRAX: High Risk
Population
Figure 2: NOGG Guidelines
CRF Based FRAX: Low Risk
<50 Years or ≥50 & No risk factors
Discharge from
Screening Program Recommend Treatment
Both CRF and BMD FRAX generate ten year probabilities of major osteoporotic fracture, which are used to designate risk of fracture
Slide by Brent Richards, McGill University, Monreal, Canada; unpublished
Copyright Prof. Dr. A.G. Uitterlinden
33. Eligible for NOGG-Based Screening
(>50 years, with at least one risk factor)
CRF Based FRAX to
calculate
10-year probability of major
osteoporotic fracture
Women with prior fragility fracture
CRF Based FRAX: Moderate Risk
CRF Based FRAX: High Risk
BMD Based FRAXBMD Based FRAX: Low Risk BMD Based FRAX: High Risk
Population
Figure 3: NOGG Guidelines with gSOS Screening Step
CRF Based FRAX: Low Risk
<50 Years or ≥50 & No risk factors
Discharge from
Screening Program Recommend Treatment
Both CRF and BMD FRAX generate ten year probabilities of major osteoporotic fracture, which are used to designate risk of
fracture. gSOS is standardized to have a mean of zero and standard deviation of one
gSOSgSOS > 0
Slide by Brent Richards, McGill University, Monreal, Canada; unpublished
Copyright Prof. Dr. A.G. Uitterlinden
34. 28 euro for GSA array
In 2016 costs of DNA analysis has gone down
Arrays are preferred in large-scale
application (compared to sequencing)
30-100x (!) cheaper
Only relevant DNA variants
Customizable
Very high throughput
Easy data analysis and automation
DTC companies prefer arrays
Less ethical issues
700,000 DNA variants on the GSA array:
GWAS, Clinical, pharmacogenetics, HLA,
forensic, mitochondrial, ancestry, blood
groups, etc.
Copyright Prof. Dr. A.G. Uitterlinden
35. 1 093 522
Europe 1 004 992
Netherlands 168 992
Canada/USA 28 209
Australia 37 219
Asia 21 952
South America 1 150
Africa 0
EU GSA consortium
Coordinating center
HuGe-F Erasmus MC
By end 2018 there will be many SNP array datasets..
Existing:
academic data 1 million samples (global)
UK Biobank 0.5 mio samples (UK)
Millions Veterans Program (MVP) 1 million samples (USA)
FinGen 0.5 mio samples (Finland)
23andme >2 mio samples (USA centric)
Avera, Kaiser Permanente 0.6 mio samples (USA)
New:
GSA sales 2016/2017/2018 >20 million samples (USA centric)
EU-GSA 1.1 million samples (global)
TOTAL ~25 million samples with SNP array data……
Copyright Prof. Dr. A.G. Uitterlinden
36. GOALL! Genotyping On ALL patients at Erasmus MC
Subscription to
GENETIC REPORT
Commercial Partners:
Illumina, BC Platforms
Pilot Projects:
-Eye disease
-Cardiovascular Disease
-Pharmacogenetics
-Breast Cancer
-Type 2 Diabetes/Obesity
-……
DNA Array
Processing:
Erasmus MC
Genomics
Core Facility
Patient’s Home
Erasmus MC
as trusted
Partner
Erasmus MC Partners:
*Interpretation/Counseling:
Internal Medicine : Complex Diseases
Clinical Genetics : Mendelian Diseases
Clinical Chemistry : Pharmacogenetics
*Patient inflow/Reporting:
Clinical departments (per disease)
Costs: < 30 euro per patient
Content: 700.000 selected variants for:
- Pharmacogenetics
- Mendelian Disease Variants
- HLA types
- Clinical (actionable) Variants
- Polygenic Risk Scores Complex Disease
- Ancestry
- etc.
Information and Consent
Regular Updates with
Risk Profile Information
*ALL patients undergo DNA
array genotyping
*Patient DNA Array results are
available BEFORE clinician sees
the patient
Copyright Prof. Dr. A.G. Uitterlinden
37. Grades of Evidence
Level Method Science disciplines
- Large scale collaborative prospective
meta-analysis of individual level data in
consortia
- Meta-analysis of published data
- >2 large studies (n > 1000 each)
- 1-3 smaller studies
- 1 small study (n<500), NO replication
- Expert Opinion…
Very Good
Not so Good
-Complex Genetics
-Physics
-Astronomy
-Sociology
-Psychology
-Medicine
Cell Biology
-The biomedical community publishes 2,5 mio papers per year
-<50% papers describe results that can be replicated (the “reproducibility crisis”)
Copyright Prof. Dr. A.G. Uitterlinden
38. *to convince yourself, colleagues, society that the observation is true and
generalizable
*because methodology in one centre is flawed:
-transformed cell lines
-wrong/mixed cell lines
-bad antibodies
-complicated/outdated genotyping method
-human error, fraud
*because effect sizes are small (e.g., GWAS, omics data)
*because the modelsystem used is not representative for humans, e.g.:
-worm/insect/mouse biology is not similar to human biology
-only one (inbred mouse) strain is used (n=1 human, and a strange one…)
-only one iPS cell line is used (n=1 human)
-a small human sample is used (cases only; an isolated population; etc.)
Replication/validation is needed (a few reasons):
> Provide replication in one and the same paper with colleagues
Copyright Prof. Dr. A.G. Uitterlinden
39. Collaboration doesn’t come easy…..
>> Donald Trump’s view on EUROPE…. ?
(From: Yanko Tsvetkov, alphadesigner.com)
Wall !!
Wall !!
Wall !!
Wall !!
Wall !!
Wall !!
Wall !!
Copyright Prof. Dr. A.G. Uitterlinden
40. A “Culture” Change in doing Research:
GLOBAL COLLABORATIONS IN COMPLEX GENETICS
Example: the “GIANT” consortium:
>2,000,000 participants…
SUNLIGHT consortium
Copyright Prof. Dr. A.G. Uitterlinden
41. - More data is better: Growth of NGS sequencing data is slow due to high
costs and complexity; GWAS by arrays/imputation grows (much) faster
- New Biology: Dozens of novel genes/pathways discovered to be involved
in disease phenotypes and risk factors
- Potential for Prediction: A still increasing part of heritability of phenotypes
is being explained
- Better Epidemiology: Mendelian Randomization is now more feasible to
analyse causality of “classic” epidemiologiccal associations
- High Impact and Exemplary: Large-scale international collaborations
allow for very robust evidence for genetic & genomic discoveries
- Populations are a bunch of individuals: opportunities for studying
“personalized/precision/stratified aspects” of biology and medicine
>> Translational Research based on these discoveries is ongoing
Population Genomics: what have we learned?
Copyright Prof. Dr. A.G. Uitterlinden
42. …..IGNORANCE CAN BE DAUNTING……EDUCATION IS IMPORTANT !!
Annual Courses organized by the Genetic Laboratory:
in 2019:
- 14th edition of “Genomics in Medicine” (Aug; ESP57; NIHES)
- 4th edition of Microbiome course (Sept; MolMed)
- 11th edition of “Genetic for Dummies” (Nov; MolMed)
- 16th edition of “SNP Course” (Nov; MolMed)
www.molmed.nl
www.nihes.nl
Copyright Prof. Dr. A.G. Uitterlinden