Polygenic and multifactorial

diseases

-key features and isolation of

responsible genes
Newcastle; 13th December 2007

Early genetics
Two camps
Followers of Mendel
Dichotomous traits
Believed that traits presented in predictable patterns of
inheritance (AD, AR, XLD, XLR & Y linked)
Biometricians
Continuous/qualitative traits eg height, weight
Not amenable to Mendels laws as variable from individual
to individual
Two camps married by demonstration that these continuous traits
could be governed by a large number of independent genes each
governed by Mendels law

Polygenic/Multifactorial inheritance
These findings form the basis of polygenic and multifactorial

inheritance and are now known to account for qualitative traits

such as height, weight, blood pressure ..
..and also for the common diseases that do not follow
Mendelian inheritance patterns and represent major health
problems such as heart disease, obesity, asthma, diabetes and
cancer

Estimated that lifetime risk of genetically influenced

common disorder in Western population is 60%

Mendelian and complex diseases

Mendelian
No of
disease
proteins

In utero
to puberty

Affecteds
per 1000
popn

Puberty of age 50

Lung
Diabetes
Heart/circulatory
Athritis/musculoskeletal

18-44

complex

45-54

Age range
Bjornnson et al 2004; TIGS Vol 20(8):350-358

Over 50

55-64

Polygenic/Multifactorial diseases
Following success in identification of the majority of single gene

disorders eg CF, HD, DMD etc

New interest in identifying genes responsible for these common

complex diseases

Improved technologies
Pharmaceutical company interest
Hope for designer medicine

Key concept of complex diseases

Multiple distinct loci interact with/without other factors including the environment
to result in end stage phenotype
Expressed in population as a continuously variable susceptibility that follows
Gaussian distribution
Effectively creates a gradient of susceptibility - phenotype presents beyond a
certain threshold

population mean

Threshold of liability

Affected individuals

Key concepts of complex disease

Familial concentration of disease without specific pattern of inheritance
Absence of clear biochemical defect resulting from single abnormal

gene
Considerable variation in severity and expression of phenotype
(between and within families)
Most affected individuals have unaffected parents
Often sex differences

Familial clustering
General population

popn mean

Threshold of liability

Affected individuals

popn mean

Siblings of affected
individuals

sib mean

Threshold of liability

Affected sibs

Recurrence risks

Affected individuals have inherited combination of high susceptibility

alleles.
Relatives share these alleles
Thus cousins, aunts, uncles etc also at higher risk than general
population
Parents with affected child have higher than average number of high

risk alleles

Recurrence risk is higher if >1 family member affected

Greater the severity of the disease the higher the recurrence risk

Different thresholds between the sexes

eg congenital pyloric
stenosis

5x more common in
boys than girls

Sex differences

Recurrence risk is greater if proband is of less commonly affected

sex
Eg Congenital pyloric stenosis
Male probands
Recurrence in brothers 3.8%
Recurrence in sisters 2.7%
Female proband
Recurrence in brothers 9.2%
Recurrence in sisters 3.8%

Cleft lip and palate

Measurement of risk
The level of risk is measured by relative risk
lR - R is the risk to a relative of affected individual compared to population risk
eg sibling risk in diabetes; lS is 15
sibling risk in autsim; lS is 145 (91-93% genetic influence)

The higher the value of l the stronger the genetic effect in disease
Relative risk for 1st degree relatives is approximately equal to square
root of population incidence
Risk decreases rapidly in remotely related individuals

Eg in schizophrenia
siblings - 10.6
first cousins - 3.6
aunts/uncles - 2.5

Genetic basis of complex disease

Classic scale of human disease
Single gene disorder
(single major locus
environment
>phenotype)

polygenic

multifactorial

(several loci

(many loci and

>phenotype)

>phenotype

Now seen as an oversimplification

Single gene disorders

single gene disorders are now known to represent disorders in which

a single major locus is necessary and sufficient to result in the

phenotype
Usually extensive allelic and non allelic heterogeneity
but where the phenotype can differ in intra and inter familial manner
due to modifier genes and environment
eg in CF

many mutations identified in the CFTR gene

good degree of correlation with CF mutation and pancreatic disease
severity of pulmonary disease shows no such correlation
now known that expression of CFTR gene is modified by a number of other
factors including variants at NOS1, HLAIII, TNFA, TGBF1, CFM etc

Polygenic diseases
digenic inheritance where defects at two loci are necessary to result in

phenotype

eg retinitis pigmentosa; peripherin-RDS, ROM1

trigenic inheritance, where defects in 3 loci/ alleles are necessary ro

result in phenotype

eg Bardet Biedl; BBS2/BBS6

oligogenic inheritance where several loci are involved in resulting

phenotype

Multifactorial diseases

Increasing no of loci involved

~18 different loci identified as being involved in diabetes
Approximately 40% familial clustering due to the HLA loci
(lS =3); other involved include INS (1.9), CTL4 (1.2)
Each locus involved is neither sufficient or necessary to result in

the phenotype
Also environmental factors

Monogenic and complex disease

Models to explain multifactorial disease

2 basic models

Common disease - common variant

(restricted polymorphism model)

Proposed that there is a small number of loci with risk alleles that are common in
the population (>1%) and each exerts a considerable genetic effect eg. APOEe4
allele in Alzheimer disease; Factor V Leiden in deep venous thrombosis

Common disease - rare variant

Suggested that there are a large number of loci with risk alleles that are rare in
the population (<1%) and each exerts little or moderate effect

In both models, different alleles at these loci may increase or

decrease risk

leads to complex patterns of susceptibility

Other factors
Environmental influence - diet, exposure to toxins, exercise etc
Epistasis - interaction between different loci;where one particular allele

at locus 1 prevents particular allele at locus 2 from manifesting its effect

Somatic changes
Epigenetics - methylation, imprinting etc

All combine to produce disease state in individual

An example of epistasis

Identification of responsible genes

Linkage analysis and association analysis are the 2 main strategies
Linkage refers to the physical coupling between 2 loci (marker locus
and disease locus
indicates close proximity of marker to disease locus
Association studies refers to the co occurrence of two variants
(particular allele of marker and phenotype)
indicates particular allele of marker (or one very close to it) is causative
in disease

Linkage vs Association

Linkage studies
Classic linkage studies require

Large mulitgenerational single family (or multiple smaller families in clear

homogeneous disease)
Defined mode of inheritance
Single locus responsible
Known penetrance
Genetic homogeneity

Clearly not the case in complex diseases

however can still be used in non parametric models under different modes of
inheritance, allowing for heterogeneity etc
thus likelihood of detecting causative locus less than in single major locus
disorders
alleles of low or moderate genetic effect unlikely to be identified

Association studies

Several methods available including

Case control series
Affected sib pair
Affected pedigree member
TDT
Each has different advantages, disadvantages and limitations
Complex statistical analysis
Can be influenced by
Sample size, selection of controls
Population stratification, admixture
Epistasis, age of disease
Problems in multiple testing
Informativeness, density of markers
Level of risk alleles effect in disease

Association studies- population based

Case-control study

Most widely applied strategy

Series of affected patients vs series of matched controls
Cases readily obtained; genotyping easy
Most prone to producing false positive results - usually due to incorrect
control population selection. Any difference in allele frequency
between groups may be due to differences between populations
(independent of disease)
Require significant numbers to adequately power study (1000s vs
100s); especially important in study of multiple variables

Case -cohort study

Cases and controls drawn from selected population under study to

investigate broad spectrum of diseases and factors
Prospective; takes longer to select sufficient numbers

Association studies in diabetes type 1

Association studies - family based methods

Affected sib pair method

Tests for increased inheritance of particular allele in sibs vs

controls
Identity by descent more powerful than identity by state
IBD requires parental testing (IBS does not)
Affected pedigree member method
Additional family members tested; similar to ASP
Discordant trios
Useful if parental samples unavailable

TDT
Transmission disequilibrium testing

Requires testing of parental and affected offspring and classifies

alleles transmitted to affected children and not transmitted
Test for Requires that parents are heterozygous
Power lost if parents are homozygous
Provides a joint test of linkage and association
Eliminates stratification effects
Can be modified to account for multi-allelic markers, multiple
siblings, missing parental data and quantitative straits

Factors influencing success of studies

Control populations (stratification)
family based controls vs matched controls
Study population
inbred populations reduce number of segregating loci and non
allelic heterogeneity
admixture; eg Latin American, African American; creates
disequilibrium that breaks down rapidly for unlinked markers;
utilised in MALD (mapping by admixture linkage disequilibrium)
Epistasis; interaction between alleles can be accounted for by
statisitical models (Markov chain Monte Carlo based methodology)

Factors influencing success of studies

Age of disease

old ancient diseases (restricted polymorphism model) have low range

of linkage disequilibrium (~3kb). Requires high density map of markers
to detect association
new diseases have high range of disequilibrium ( 10kb). Low density
scans required but low power to detect

Genetic effects of risk allele

Few loci exerting considerable effect

Power to detect reduces with increasing no of loci

Informativeness of markers

power to detect decreases with reduced heterozygosity

Inference of linear distance

Distance between marker and disease not easy to predict due to non
linear relationship between LD and distance below 60kb

Identification of risk alleles

Linkage/association studies to locate regions of interest
Finer mapping of regions
Sequence analysis of candidate genes within interval
Numerous sequence variants likely to be present
Role of identified variants? Combination of variants?? Logistical

challenge
Functional tests of candidates; cellular testing, knock out/in
models etc

Polygenic/Multifactorial diseases
Encompass latter end of spectrum of human disease
Result from combination of numerous loci each of which is

neither sufficient nor necessary to result in disease

Identified by combination of linkage and association studies
Statistical and logistical issues

Selected references

Bjornsson et al (2004) TIGS Vol 20(8):350-358

Weeks and Lathrop (1995) TIGS Vol 11(12); 513-519
Smith & OBrien (2005) Nature Review Genetics online pub 12 July; 1-10
Cardon & Bell (2001) Nature Review Genetics Vol 2; 91-99
Laird & Lange (2006) Nature Review Genetics Vol 7; 385-394
Wright et al (1999) Nature Genetics Vol 23; 397-404
Badano & Katsanis (2002) Nature Review Genetics Vol 3; 779-789
Wright et al (2003) TIGS Vol 19 (2); 97-106
Antonarakis & Beckman (2006) Vol 7; 277-282
Lanpher et al (2006) Nature Review Genetics Vol 7; 449-460
Concannon et al (2005) Diabetes Vol 54; 2995-3001
& Strachan & Read