Healthcare Management

Chapter 14
Molecular and Genetic Epidemiology

Learning Objectives
Differentiate between molecular and genetic epidemiology
Describe principles of inheritance and sources of genetic variation
Define epidemiologic approaches for the identification of genetic components to disease

Peeking into the “Black Box”
Many risk factors can be quantified through questionnaires, records, and easily measured attributes (such as blood pressure and anthropometrics).
The biological mechanism(s) through which these factors influence disease is not always apparent (i.e., a “black box”).

Value of Mechanistic Insight
Biologic plausibility is a criterion for causality.
Linking lifestyle risk factors with measures of biologic effect strengthens interpretations of causality.
This linkage, in turn, provides stronger support for interventions.

Why Distinguish Between Molecular and Genetic Epidemiology?
The basic tenets and principles of molecular and genetic epidemiology are the same.
However, there are specific features regarding design, analysis and interpretation inherent in the latter.

3

Definition of Genetic Epidemiology
A discipline that seeks to unravel the role of genetic factors and their interactions with environmental factors in the etiology of diseases, using family and population study approaches.

4

Key Aspects of This Definition
Inherited susceptibility does not mean inherited disease–environment matters!
When families are studied, the observations (study subjects) are no longer independent.
This dependence requires special considerations for the analysis of data.

5

Genetic Epidemiology is a Method to Answer:
Does a disease cluster in families?
If so, is that clustering likely a result of shared non-genetic risk factors?
If the clustering is not accounted for by shared lifestyle or common environment, is the pattern of disease consistent with inherited effects?
If so, where is the putative gene?

6

What Diseases or Risk Factors Cluster in Families?
Heart disease
Various cancers
Alcoholism
Others

Epidemiologic Assessment of Clustering
Case-control study
Comparison of the frequency of a positive family history
Expectation under genetic influence

Clustering of “Non-Genetic” Exposures in Families
Employment (e.g., several family members with medical degrees)
Radon from soil
Religious preferences
Lead in paint
Others?

7

Major Point of This Section
You cannot tell easily whether clustering of a risk factor or disease within a family is due to genetics, culture, or shared environment (including social or political factors).
Clustering within a family will also occur simply due to bad luck!

8

Other Correlates of Family History
Large family size
Age of relatives (for an age-related disease)
Gender distribution (consider testicular cancer, prostate disease, ovarian cysts)

Analysis Approach
Model Y (case/control status) = established risk factors.
Add family history variable to denote “genetic” influence (i.e., share genes with an individual who has the outcome of interest).

Analysis Issues
Try to compare (and control if necessary) differences between cases and controls with regard to size of family.
Not easy to adjust for age of family members or their risk factors.
What types of data can you ask your cases and controls to provide about their relatives?

Motivation for Case-Control Family Studies
To rule out influence of shared environment, family size differences, and age on differences in the frequency of family history between cases and controls
Need to enumerate the relatives of cases and controls, and determine the disease status and risk factor profile for each relative

Conduct of Family Studies
Ascertain “probands” (index cases).
Define family (siblings? children? parents? grandparents?)
Invite family members to participate
Collect data (and, typically, biological samples)

How to Select Control Families
Must decide how to identify controls
From spouse’s side of proband’s family?
Or select a random sample from the population?
Will controls be motivated to participate?
Must take HIPAA rules into account

Analysis Issues
Exclude the index cases and controls
Model disease (or behavior) of interest based on age, sex, known risk factors
Evaluate evidence for genetic effect through statistical significance of variable(s) that indicate “relationship to index case”

Analysis Issues (cont’d)
Simplest “genetic” variable (1 if relative of case, 0 if relative of control)
Can also construct indicator variables to designate type of relative (parent, sibling, more distant relative)
If not significant after including other risk factors, then no evidence for genetic influence

Evidence of Genetic Influence, so far….
Cases are more likely to have a family history of disease than controls.
The excess risk to relatives is not accounted for by age, sex, and other risk factors.
What does that tell us about the underlying genetic influence? (nothing)

Other Approaches to Identify Genetic Influences
Twin studies
Segregation analysis
Linkage analysis

Twin Studies
A “natural experiment” of sorts
Monozygotic (MZ) twins are genetically identical.
Dizygotic (DZ) twins share, on average, the same proportion of genes as siblings.
Greater concordance (for dichotomous traits) or correlation (for continuous traits) for MZ than DZ twins is evidence of a genetic influence.

9

Linkage Analysis
One way to distinguish cultural inheritance from genetic inheritance is to track a region of our DNA that is transmitted from parents to offspring in the same manner as the disease/outcome of interest.
This procedure works well for diseases that follow simple rules of inheritance (e.g., autosomal dominant or recessive).

10

Segregation Analysis
Historically, linkage analysis required knowledge of the mode of transmission of the putative gene [dominant versus recessive, allele frequency, lifetime or age-specific risk (penetrance)].
Segregation analysis has been used to estimate these parameters.

Genetic Epidemiology of Complex Diseases
“Complex diseases” are ones for which the genetic influence may be modest and environmental factors contribute to disease risk.
Segregation analysis is not typically done for “complex diseases.”
Modern approaches ignore models of inheritance (non-parametric methods).

Use of Epidemiology to Understand Genetic Variation
The methods of genetic epidemiology have been applied historically to identify genes.
Typically, epidemiologists are not interested in mapping genes, but rather in figuring out how genes interact with environment to influence disease risk and outcome.

Molecular Epidemiology
Related individuals are not necessarily required for studies of the association of genetic variation with risk of disease.
Both cohort and case-control designs can be used.
Because genetic code (germline DNA) is unchanged since conception, one readily can employ retrospective designs.

Common Strategies for Genetic Marker Selection
Genome-wide approach with anonymous DNA markers (1,000,000 SNPs on a chip)
SNPs or simple tandem repeat markers in “candidate” genes based on a priori knowledge about presumed function
SNPs in candidate genes with known functional effect on level or activity of protein product

Primer on Single Nucleotide Polymorphisms (SNPs)
Because of our redundant genetic code, some SNPs will not alter the encoded amino acid (e.g., GGA, GGG, GGT and GGC all encode proline).
SNPs that change an amino acid may not necessarily lead to change in function of transcribed protein.

More on SNPs
SNPs that don’t change an amino acid may still lead to alternate splicing of the transcript (and therefore be functionally important).
SNPs in promoter region may influence level of protein product–not activity (and therefore be biologically significant).
SNPs in non-coding regions may still have functional effect.

Caveats About SNP Studies
If you’re interested in gene x environment interactions–best to focus on SNPs with known functional effect.
Human biology is complex: are alterations in one component of a pathway compensated for by another?
Most SNPs are likely to be modest risk factors–requiring large sample sizes to determine statistically significant association.

Realistic Expectations
Almost every gene is modified after translation into protein (e.g., glycosylation, acetylation, methylation).
Thus, the correlation between DNA sequence and protein is far from perfect.
Most GWAS “hits” are in “gene deserts.”
May be necessary to examine multiple SNPs within a gene and several genes within a pathway.

Molecular Epidemiology – Beyond Genetics
Biomarkers of exposure and disease extend beyond DNA.
Viral or bacterial load
Morphometric analysis of tissues/cells
Hormone or lipid levels in blood or urine
Other examples?

Conclusion
Molecular and genetic epidemiology represent specialty areas of expertise.
These specialty areas utilize and apply advances in molecular biology and molecular genetics of disease to:
Unravel disease etiology.
Enable novel approaches for early detection.
Inform more effective interventions by targeting those at greatest risk.