EPB PHC 6000 EPIDEMIOLOGY FALL, 1997

1
Lecture Presentation Topic Epidemiological and
statistical methods for analysis of
gene-environment interaction study By Victor
Okoh  PHC 6921- Environmental Health
Toxicology Seminar Instructor Dr. D. Roy
2

• Lecture Objective
• Understand characteristics, uses, strengths, and
  limitations of genetic epidemiology study
• 2. Understand the concept of G/E
• interaction.
• 3. Understand characteristics and interpret
  results from gene-environment studies
• 4. Distinguish between statistical and biological
  gene-environment interaction, including
  assessment of when interaction is present

3
Background and Definitions
4
Background
Definitions Gene Particular segment of DNA
molecule on a chromosome that determines the
nature of an inherited trait. Locus Site or
location on a chromosome occupied by a
gene. Allele One or two or more alternate
forms of a gene that occurs at the same locus.
5
Background
Definitions Genotype Genetic constitution
of an individual, often in reference to a
particular trait. Phenotype Realized expression
of the genotype. In epidemiology, genotype is
modified by the environment to affect the
phenotype. Proband The individual in a family
that brings attention to the investigator (e.g.
the case or control proband of a set of twins).
6
Background
As stated in previous class, MOST chronic
diseases like cancer have multi-factorial
etiololgies At least, the individual must be
genetically susceptible to the disorder AND
There must be at least some interaction with
the environment.
7
Background
Classic Example of Gene-Environment
Interaction --- Persons with HLA-B27 are
approximately 90 times more likely to develop
ankylosing spondylitis than persons without
HLA-B27 BUT --- Only 10 of all persons
with HLA-B27 will develop ankylosing spondylitis.
8
Background
Other diseases with GE interaction
Cancer Asthma Birth Defects
Cardiovascular diseases Chronic neurological
diseases Parkinsons disease Alzheimers
disease Lead poisoning Pesticides Immune
dysfunction
9
Study Designs in Genetic Epidemiology
10
Study Designs in Genetic Epidemiology
1) Family Studies 2) Twin Studies 3) Adoption
Studies 4) Migrant Studies 5) Genetic Marker
Studies --- Gene/Environment interactions
Traditional
Recent
11
Overall, basic strategy of these designs are
to a) Hold environment constant ? allow
genetic factors to vary b) Hold genetics
constant ? allow environmental factors to vary
12
In general, the traditional studies designs are
used to assess the relative contribution of
genetics to disease occurrence, whereas the
recent studies attempt to identify genetic
factors that play a causal role in disease
development.
13
Family Study Design
vary genetics, keep environment
constant 1) Identify individual with
particular disorder (case-proband) and compare
that with rates of the disorder in relatives of
the proband. 2) Identify individual without the
disorder of interest (control-proband)
determine rates of the disorder in
relatives of the proband. 3) Calculate
prevalence ratio between relatives of case and
relatives of control proband.
14
Family Studies
vary genetics, keep environment
constant 1) Identify individual with
particular disorder (case-proband) determine
rates of the disorder in relatives of the
proband. 2) Identify individual without the
disorder of interest (control-proband)
determine rates of the disorder in
relatives of the proband. 3) Calculate
prevalence ratio between relatives of case and
relatives of control proband.
15
Family Studies
vary genetics, keep environment
constant 1) Identify individual with
particular disorder (case-proband) determine
rates of the disorder in relatives of the
proband. 2) Identify individual without the
disorder of interest (control-proband)
determine rates of the disorder in
relatives of the proband. 3) Calculate
prevalence ratio between relatives of case and
relatives of control. PR gt 1.0 indicates suggest
genetic etiology of disease
16
Family Studies
Case
Control
Proband
Relatives
Prevalence ratio (PR) (3 / 5) / (1 / 5) 3.0
17
Family Studies
Limitations of Family Studies a) Important
environmental factors (e.g. SES, social support,
etc.) also tend to be familial. b) There may
be tendency for those with particular disorders
to mate preferentially with those who have
similar disorders may lead to overestimate of
genetic effect.
18
Twin Study Design
vary genetics, keep environment
constant 1) Identify one twin with disorder of
interest. 2) Assess whether other twin has the
disorder of interest. 3) Calculate the
concordance rate (CR) 4) Compare concordance
rates between monozygotic (MZ) and dyzogotic
(DZ) twins. 5) To support genetic etiology,
concordance rates for MZ twins should exceed rate
for DZ twins.
19
Types of Twin Study Designs
Pairwise concordance is used when the probability
of selecting both cases as twins is very low
(e.g. selecting one proband case from each
city). Probandwise concordance is used when the
probability of selecting both case probands as
twins is not rare (e.g. selecting every third
person as a case from a local registry of disease
cases). Note The probandwise method usually
results in somewhat higher concordance rates.
20
Twin Studies
Concordance rate (pairwise method)
twin pairs concordant CRPW
------------------------------------------
total twin pairs Concordance rate
(probandwise method) 2C1 C2
CRPB -------------------------------------------
-------------- 2C1 C2 D Where
C1 Concordant pairs both twins
independently ascertained C2
Concordant pairs only one twin independently
ascert. D Number of discordant pairs
21
Twin Studies
MZ Twins
DZ Twins
Twin 1
Twin 1
Twin 2
Twin 2
CRMZ A / (A B C)
CRDZ A / (A B C)
Genetic contribution CRMZ / CRDZ
Note Discordance rate (B C) / (A B C)
22
Twin Studies
MZ Twins
DZ Twins
Twin 1
Twin 1
Twin 2
Twin 2
CRMZ A / (A B C)
CRDZ A / (A B C)
Genetic contribution CRMZ / CRDZ
CRMZ (30 / 140) 0.2143
CRDZ (32 / 288) 0.1111
CRMZ / CRDZ 0.2143 / 0.1111 1.93
23
Twin Studies
MZ Twins
DZ Twins
Twin 1
Twin 1
Twin 2
Twin 2
CRMZ / CRDZ 0.2143 / 0.1111 1.93
The MZ/DZ ratio of 1.93 suggests that genetics
have a strong influence on development of the
disorder.
24
Twin Studies

• Some Other Uses of Twin Studies
• MZ twins reared apart give unique (but rare)
  opportunity to study influence of genetics in the
  absence of shared environmental factors.

25
Twin Studies

• Important Features of Twin Studies
• Twins constitute about 1.8 of adult population.
  Use of twin registries allows study in the
  community rather than the hospital (avoids
  treatment seeking bias and lack of
  generalizability).
• When 2 or more disorders are studied in twins,
  can estimate comorbidity due to shared genetic
  and shared environmental factors.
• Keep in mind that the equal environment
  assumption between MZ and DZ twins may be suspect.

26
Adoption Study Design
Vary environment, keep genetics constant Based
on premise that if genes are important, disorder
transmission should occur in the biologic family
but not adoptive family. If environment is
important, disorder transmission should occur in
the adoptive rather than biologic family.
27
Three major adoption study designs Parent-as-pro
band design Adoptee-as-proband
design Cross-fostering design (most powerful)
28
Adoption Studies

• Parent-as-proband design
• Compares rate of illness in the adopted
  offspring parents with and without the disorder
  of interest.
• If genetic factors are important, rates of
  illness should be higher in adopted children of
  ill parents compared with adopted children of
  well parents.

29
Adoption Studies Parent-as-Proband
Parent 1
Parent 2
Biological Parents
Adopted Offspring
Adopted Offspring
Suggests genetic heritability
30
Adoption Studies
b) Adoptee-as-proband design Start with ill and
well adoptees, and examine rates of illness in
both biologic and adoptive relatives. If
biologic relatives show higher rates of illness
than adoptive relatives, suggests a genetic
component. However, if adoptive relatives show
higher rates of illness, environmental hypothesis
gains support.
31
Adoption Studies Adoptee-as-Proband
Adoptee 1
Adoptee 2
Biologic Relatives
Adoptive Relatives
Biologic Relatives
Adoptive Relatives
(Suggests genetic component)
32
Adoption Studies

• Cross-fostering design Compares rates of
  illness for 2 groups of adoptees
• Group 1 Adoptee has well biologic parents,
  raised by ill adoptive parents.
• Group 2 Adoptee has ill biologic parents,
  raised by well adoptive parents.
• Higher rates of illness in Group 1 suggests
• non-genetic mode of transmission.

33
Adoption Studies Cross Fostering Design
Biologic
Adoptive
Biologic
Adoptive
Parent of Adoptees
Adoptees
Adoptees
(Suggests environmental component)
34
Adoption Studies

• Limitations of Adoption Studies
• Limited generalizability since adoptees and their
  families not representative of the general
  population.
• Adoptees are often at greater risk of illness
  than non-adopted children (e.g psychiatric
  disorders).
• Parents of adopted children are known to have
  higher rates of some disorders (e.g.
  psychopathology).
• May be difficult to find sample of adoptees
  separated from parents at birth (e.g. biologic
  relationship contaminated by environment of
  biologic parents).

35
Migrant Studies

• vary environment, keep genetics constant
• Evaluate incidence of disorder among
  ethnically-similar individuals living in native
  environment.
• Evaluate incidence of disorder among comparable
  group of ethnically-similar individuals who have
  migrated to a new environment that has much
  higher or lower rates of the disorder.
• As the rates of disorder among migrants approach
  those observed in the new environment, evidence
  for environmental influence increases.

36
Migrant Studies
Examples of Findings from Migrant Studies
With exception of stomach cancer, Japan
historically has had much lower rates of cancer
than the U.S. BUT, after just 2 generations,
Japanese migrants to the U.S. have nearly assumed
the same rates of cancer (including lower rates
of stomach cancer). European immigrants (high
latitude) who migrate to Israel (low latitude) at
a young age experience a low incidence of
multiple sclerosis. However, if the migration
occurs after age 14, the relevant environmental
exposure may have already occurred.
37
Migrant Studies
Some Limitations of Migrant Studies
--- Migrants are highly selected. --- Age at
migration varies prior causal exposure in the
native environment may have already
occurred. --- Retainment of cultures and
lifestyle of original environment make it hard
to measure effects of new environment
38
Genetic Marker Studies

• vary environment, keep genetics constant
• Genetic markers are a measurable human trait
• controlled by a single gene with known
• chromosomal location.
• Must be polymorphic with at least two alleles
• having a gene frequency of at least 1 percen1)
• Genetic Marker studies are used examine
• links between genes and specific disorders
• or traits.

39
Genetic Marker Studies

• Markers Used to Study Allele/Disease Associations
• --- Blood groups
• --- Human leukocyte antigens (HLAs)
• --- Protein polymorphisms
• Analysis of DNA polymorphisms
• --- Allelic variants of genes
• --- Restriction fragment length polymorphisms
• --- Variable tandem repeats
• Note The marker may or may not be causally
  involved in development of the disease.

40
Genetic Marker Studies

• Two types of genetic marker studies
• a) Association studies
• b) Linkage studies

41
Genetic Marker Studies
Association studies Select disease cases and
controls using case-control design.
THEN Compare the frequency of the
genetic marker of interest (e.g. specific
allele) between cases and controls.
42
Genetic Marker Studies
a) Association studies
43
Genetic Marker Studies
b) Linkage studies Assess the association of
general markers and disease within families.
--- Linkage is defined as the tendency of
genes to be inherited together as a result of
their location on the same chromosome. ---
Genetic Linkage is measured by LOD score
(logarithm of the odds)
44
Example of Linkage Analysis
45
Genetic Marker Studies
Linkage (background) --- Genes co-exist on
chromosomes. 22 pairs of non-sex chromosomes
(autosomes) 1 pair of sex chromosomes (XX F,
XY M) --- In meiosis, 2 pairs of chromosomes
segregate one chromosome of each pair is
transmitted to gamete --- Chromosomes segregate
independently. --- However, chromosomes do not
always stay intact crossing over
(recombination) occurs. --- Frequency of
crossover between 2 loci on same chromosome
depends on their distance apart.
46
Genetic Marker Studies
What is LOD score? a statistical estimate of
whether two sites of genes (loci) are likely to
lie near each other on a chromosome and are
therefore likely to be inherited together as a
package
47
Genetic Marker Studies

• Lod Score Method --- Assumptions
• 2 traits of interest are inherited as simple,
  single locus Mendelian traits.
• Relations between genotype and phenotype are
  known (e.g. dominance and penetrance).
• Gene frequencies of various alleles are known.
• For a family to be informative for linkage, one
  parent must be heterozygous for both traits of
  interest (e.g. Aa, Bb).

48
Genetic Marker Studies
The Lod Score is the log of the odds ratio.
For monogenetic diseases that have
Mendelian patterns of inheritance gt
3 evidence of linkage lt -2 evidence against
linkage gt -2 to lt 3 inconclusive However,
Lod Scores required to confirm or reject linkage
for complex disorders (e.g. psychiatric
disorders) need to be of greater magnitude due to
complex patterns of inheritance.
49
Summary
These are a few of the study designs used by
Epidemiologist and biostatisticians to study the
effects of Gene/environment on disease
etiology. The unraveling of the human genome has
enabled epidemiologist and biostatisticians to
further our understanding of how gene and
environment work in concert to affect etiology
of human disease
50
Credits
Dr Sandberg of School of Public
Health.University of Pittsburgh