Torben Hansen

1
Torben Hansen
Genetics of Diabetes STAR Epidemiology
Course Mumbai, February 2003
2
Why find genes?

• Molecular understanding

New targets
Improved dissection of environmental factors
Individual genetic prediction
Improved nosological classification
pharmacogenomics
New drugs
Better prevention
Individualised treatment
3
Genetics the inherited contribution to
phenotypic variation
agaatttcat atT/Cgtg gaagaggaca
3.2 billion letters of human DNA
4
LIKE TRUE LOVE,TRUE DIABETES GENES REMAIN HARD TO
FIND
5
Empirical risks for Type 1 diabetes

• Background population 0,4
• Average risk for siblings 6
• One parent with Type 1 diabetes 2-5
• Both parents affected 5-20
• HLA-identical siblings 12
• Monozygotic twins 35-70

6
Empirical risks for Type 2 diabetes

• Background population 10
• Average risk for siblings 30-40
• One parent with Type 1 diabetes 30-40
• Both parents affected 50-80
• Monozygotic twins 50-90

7
Familial Clustering
6
risk to siblings
l
15


population prevalence
0.4
T1D
30-40
risk to siblings
l

3-4

population prevalence
10
T2D
8
The mode of inheritance of IDDM is complex
(read unknown)
A pure multiplicatory model is most often assumed
when evaluating data from genome screenings
l S a x b x c x d x e
9
Possible Models for theGenetic Basis of Type 2
Diabetes
10
The Human Chromosomes

• 46 chromosomes
• - 22 pairs of autosomes
• - 1 pair of sex chromosomes
• Total length of the humane genome
• - 3.3 x 109 basepairs in
• the haploide genome (physical)
• - 3.000 cM (genetic)
• - 1 cM 1 Mb

Two loci which show 1 recombination are
defined as being 1 centimorgan (cM) apart on a
genetic map
11
(No Transcript)
12
Alternative splicing A single gene can produce
multiple related proteins or isoforms, by means
of alternative splicing
N Engl J Med 347 1512-1520 2002
13
Genetic dissection of diabetes

• Linkage approach
• Large families
• Affected sib-pairs
• Quantitative traits
• Candidate gene approach
• Differential RNA/protein expression
• Animal models
• Bioinformatics

14
Genome Scan Concept

• Scan the entire genome with a dense collection
  of genetic
• markers
• Calculate appropriate linkage statistics at
  each posi-
• tion along the genome
• Identify regions which show a significant
  deviation from what
• would be expected under independent
  assortment
• Clone the disease-contributing gene from the
  linked region

15
The inheritance of diabetes

• Mendelian

• Complex

1
2
1
2
3
1
2
3
4
5
6
7
8
1
2
3
4
1
16
(No Transcript)
17
Maturity Onset Diabetesof the Young (MODY)

• Monogenic form of type 2 diabetes
• Autosomal dominant mode of inheritance
• Type 2 diabetes appears at a low age
• Impaired insulin secretion is a major phenotypic
  trait
• Different subtypes due to mutations in different
  genes

18
MODY subtypes

• MODY1 Hepatocyte Nuclear Factor-4a
• MODY2 Glucokinase
• MODY3 Hepatocyte Nuclear Factor-1a
• MODY4 Insulin Promotor Factor-1
• MODY5 Hepatocyte Nuclear Factor-1b
• MODY6 NeroD1

19
Clinical characteristics of different MODY forms
MODY1 MODY2 MODY3 MODY4 MODY5 MODY6 HNF-4
a Glucokinase HNF-1a IPF-1 HNF-1b NEUROD Locus
20q 7p 12q 13q 17q 2q Fasting
hyperglycemia 0- 0- 0- ?
0- ? Postprandial hyperglycemia - 0-
- ? - ? Minimum age at
diagnosis 7-9 years 1 year 5 years 1 family 4
families 2 families Need for insulin
therapy 30 2 30 ? ? 50 Late diabetic
complications common rare common ? ? ? Patophysi
ology beta-cell beta-cell beta-cell beta-cell beta
-cell beta-cell Prevalence of MODY 5 10-15 60-7
5 rare rare rare Non-diabetes related ?
s-trig reduced ? renal pancreatic renal
cysts, Features birth weight threshold, agenesi p
roteinurea sulf.urea in renal
failure sensitivity homozygotes
20
ARE HLA-X/X IDDM PATIENTS MISCLASSIFIED MODY3
CASES?
39 unrelated Danish patients (age at onset 12.7 Y
(3-20), GAD65 and/or IA-2 antibody positive 36)
screened for HNF-1a gene mutations

• 4 positive (heterozygous, one GLU48LYS, two
  CYS241GLY,
• one PRO291fsdelA)
• All have IDDM in 3 or 4 generations
• All patients are autoantibody negative

Conclusion 10 of HLA-X/X antibody negative
IDDM patients may be MODY3 (i.e. 1 of all
IDDM patients).
Møller et al. Diabetologia 411528 (1998)
21
Sharing identical by descent
2
1
0
Expected ratio
0.25
0.5
0.25
22
Statistics
Replication study MLS gt 1.2 nominal
p-value lt 0.01
All regions with a nominal p-value of p 0.05
encountered in a complete genome scan are worth
reporting - without any claims of linkage
23
Comparisons
24
 



 

25
Genetic predisposition to IDDM
Locus
Chromosome location
6p2111p1515q2611q136q2518q212q316q25-q273q
21-q2510p11.2-q11.214q24.3-q312q332q346q21 Xq
7p
IDDM1IDDM2IDDM3IDDM4IDDM5IDDM6IDDM7IDDM8ID
DM9IDDM10IDDM11IDDM12IDDM13IDDM15 DXS1068GCK
Davies 1994 and Hashimoto 1994
26
Scepticism
We finished the DM genome map - now we cant
figure out how to fold it !
27
Differences between populations

• Differences in design
• Type 1 and 2 statistical errors
• Reflect true heterogeneity

Heterogeneity between populations

• Epistasis
• Phenotype vs. genotype
• Gene-environment interaction

28
Replication studies in complex disorders
Study 1 Identifies 4 of 20 polygenes Replicatio
n study 4/20 x 4/20 1/25 4 chance for
identical result
29
Gene-enviroment interaction
If your child looks like you its genetics. If
it looks like the neighbour its environment.
30
A Model for the Natural History of Type 2 Diabetes
Normal glucose tolerance
Impaired glucose tolerance
Non-diagnosed type 2 diabetes
Type 2 diabetes
0
30
45
60
Age (yr)
Diagnosis

• Genes predisposing to
• Insulin resistance
• Insulin deficiency
• Obesity
• Intra-uterine growth
• retardation

30-50 of all cases have late diabetic complicati
ons at the time of diagnosis

• Environmental factors
• Acquired obesity
• Sedentary life
• Smoking
• Exogenous toxins

31
Studies of familiality andquantitative trait
loci (QTLs) in glucose tolerant offspring of T2D

• Obesity, low insulin sensitivity, low insulin
• secretion and low birth weight predict risk
• of T2D among Caucasians
• Q
• How muh of each of these predictors is familial?

32
Advantages of studying quantitative traits (QTs)

• Reduction of genetic heterogeneity by analysis of
  main pre-diabetic traits, i.e., insulin
  resistance and insulin deficiency
• Families available for genetic studies
• Categorical diagnostic states are not required

33
Steno QTL-Study ofglucose tolerant relatives of
T2D

• 61 Danish Caucasian T2D probands with 4 or more
  
• glucose tolerant offspring. A total of 246
  subjects
• T2D probands were diagnosed after age 45 yrs
• Patients with autoimmune diabetes (ICA or
  antiGAD positive), MODY or MIDD were excluded
• Typical family structure

Proband
34
Major T2D-related quantitative traits in 246
glucose tolerant offspring of one T2D proband

• Glucose tolerance
• 2 h post-OGTT glucose
• AUC glucose
• Fasting plasma glucose
• Kg (IVGTT)

• Insulin sensitivity
• Bergman
• HOMA
• OGTT-estimates

• Insulin secretion
• First phase during IVGTT
• Responses during OGTT
• Tolbutamide induced
• secretion

• Glucose effectiveness
• Sg

• Lipid metabolism
• s-triglycerides
• s-FFA

• Body- fat and composition
• Fat mass
• Waist circumference
• BMI

• Physical fitness
• VO2-max

35
Familiality of quantitative traits

• Familiality (h²) ratio of the additive genetic
  variance to the total phenotypic variance using a
  variance component model and adjusting for BMI,
  age and gender
• A polygenic model was assumed. Additive genetic
  variance also includes the impact of shared
  environment
• No dominance effects were found

36
Familiality of estimates of body fat, body mass
index and body composition
37
Quantitative trait chromosomal loci (QTLs) in
glucose tolerant relatives of T2D

• Obesity, low insulin sensitivity, low insulin
• secretion and low birth weight predict risk of
  T2D
• Q
• Can QTLs of the predictors be identified?

38
Random genome mapping of quantitative trait loci

• ABIs Linkage Mapping Set
• 340 microsatellite markers
• 10 cM spacing
• Average heterozygosity 0.8
• Variance component model

39
Obesity-related QTLs
Trait Chr Pos (cM)
LOD BMI 8 50 3.3 BMI 12
76 3.1 body fat 6 66 2.9
40
Chromosome 8 peaks
Acute insulin response
BMI
Max LOD score
Max LOD score
3
3
2
2
1
1
0
0
0
50
0
0
50
0
100
150
100
150
Map position (cM)
Map position (cM)
41
Results - chromosome 8
LOD 2.55 _at_ 42cM 9 of genome scan replicates
42
Discovery
43
Genome scan to gene
0.3 genome 10,000,000 bases 100 genes 30,000
common variants gt5,000 perigenic variants
10cM
lod
44
Triangulation
Positional candidates
Large-scale genetic epidemiology
45
The candidate gene approach
Gitte Andersen Steno Diabetes Center
46
Genetics of diabetes mellitus
Unkown type 2 diabetes genes ( 70 )
Wolfram (lt 1 )
MIDD (lt 1 )
Insulin receptor (lt 1 )
MODY ( 4 )
Unkown LADA genes ( 10 )
Unkown type 1 diabetes genes ( 15 )
47
Calpain-10
HNF-1a
IRS-1
IPF-1
PGC-1
InsVNTR
B2AR and B3AR
SUR1/Kir6.2
PPARg
GCK
InsVNTR
B3AR and PPARg
PTP-1B
Insulin action
Insulin secretion
Obesity
Birth weight
Gene-gene and gene-environment interactions
Environment
Subsets of polygenic T2D
48
Future studies Complementary attacks

• Whole genome association studies based upon
  haplotype mapping - will it be possible?

49
Association studies and common complex diseases

• Advantages
• Most genetic variation is common
• Greatest statistical power
• Easy to collect large samples of cases
• Genome sequence and SNPs increasingly available

50
Disease association direct and indirect
Direct Causative SNP
Indirect Ancestral Segment

• 10million SNPs

51
Working definition of a haplotype block

• A region over which ancestral segments have been
  inherited without historical recombination

Gabriel et al, Science, 2002
52
(No Transcript)
53
Genome coverage by haplotype blocks

• European and Asian subjects
• Mean size of 22 kb
• Estimate 80 of the genome in blocks
• gt 10kb
• Gabriel et al, Science, 2002

54
Random genome association studies in common
diseases

• Public haplotype block maps will be
• available in 2004/2005
• High-through-put and cheaper SNP genotyping is
  being developed
• In a few years it will be possible to do random
  genome association studies applying about 250.000
  haplotype blocks

55
SNPs and pharmacogenetics
What is an SNP?
Different people can have a different
nucleotide or base at a given location on a
chromosome
What is an SNP map?
Location of SNPs on human DNA
Human DNA
How can an SNP map be used to predict medicine
response?
Section of SNP genotype profile
Patiants with efficacy in clinical trials
Patiants without efficacy in clinical trials
Predictive of efficacy
Predictive of no efficacy
Nature, v405857-865
56
Development of a pharmacogenetic medicine
response profile
Abbreviated SNP linkage disequilibrium profile
for efficacy and common adverse events
Used to select patients for phase III clinical
trials
Abbreviated SNP linkage disequilibrium profile
for serious, rare adverse events
Comprehensive medicine response profile to
predict efficacy and adverse events
Nature, v405857-865
57
Genetic testing in the future
Genetic testing
Disease genetics Disease prognostics/diagnostics
Pharmacogenetics Medicine response profiles
Utility
Rare mendelian diseases causal genes
Common complex diseases susceptibility genes
Genes for drug metabolism and/or action
SNP profiles for drug metabolism and/or action
What is tested
New disease insights and future medicines and/or
prevention
Optimal medicine response
Benefits
Nature, v405857-865
58
Why find genes?

• Molecular understanding

New targets
Improved dissection of environmental factors
Individual genetic prediction
Improved nosological classification
pharmacogenomics
New drugs
Better prevention
Individualised treatment