Genetics of Complex Disease

Bert Gold, Ph.D., F.A.C.M.G

Complex versus Mendelian Trait

• Complex trait suggests the involvement of:

– Multiple loci and/or

– Environmental effects

Characteristics of complex traits

• May be difficult to determine

– May require measurements– Instruments– Phenotype details may be expensive to

collect– Affected may be well defined, the

unaffected may not

Mendelian trait suggests

• Single locus– May have some genetic heterogeneity– Or reduced penetrance but,– Strong genotype-phenotype correlation

Complex traits are common

• They may cluster in families, but not demonstrate an obvious Mendelian pattern

• Phenotypes may derive from multiple genes and environment acting in concert

• e.g. depression, schizophrenia, cardiovascular disease, dyslexia.

• Complex traits have substantial impact on public health.

• Mendelian traits are rare.

How do you know a complex phenotype is genetic?

• Mendelian disorders– Recognizable inheritance patterns– Phenotypic expression is highly correlated with

genotype at the disease locus

• Complex disorders– Qualitative or Quantitative Traits– Qualitative

• Presence or Absence– Breast Cancer– Bipolar Disease

How do you know a complex phenotype is genetic?

• In complex traits, absence unaffected• Phenotypes are defined as above a

threshold value• Obesity => BMI 27 kg/m• Diabetes=> Fasting plasma glucose >

126 mg• Threshold for affection status may be

somewhat arbitrary.

Quantitative Traits

• Measurements– Height, Weight, BMI– Fasting plasma glucose– Blood pressure

• Scales– Mini-mental status examination (MMSE)– Autism Diagnostic Interview – Revised (ADI-R)– Risk Scores– Evidence for a genetic component?

More about Complex Traits

• Usually have no distinct inheritance pattern

• High frequency in the general population complicates determining who carries the genetic liability within a family

• Phenotypic expression modified by the genotype of several loci and environmental factors.

Genetic Dissection of a complex disease

• Requires– Clinical definitions– Study design perspectives– Statistical approaches– Molecular approaches– Social, legal, and ethical issues

discussions and clearances

The Literature Review

• How was the disease defined/diagnosed in previous studies?

• How does this compare with your clinical methods?

• Were individuals of all ages studied or only one age group?

• What races were studied?• What environmental factors were considered?• Significant deviations from your own definition

of phenotype and study may necessitate you repeat these analyses.

Approaches to Determining the Genetic Component of a

Disease

• First, establish evidence of a familial component

• Second, determine the cause of familial aggregation

• Third, identify the specific genetic factors involved

Methodology for Establishing the Genetic Component

• Segregation analysis

– Computer/labor intensive statistical tool (SAGE, PAP) used to examine inheritance patterns for a disease.

– By comparing different models, you can determine which inheritance model provides the “best fit” to your data

Segregation Model

• Advantages:

– When successful, it provides a genetic model for linkage analysis.

– Confirms that major gene is present before investing in expensive genomic screening.

• Simple example assuming autosomal dominant inheritance

Segregation Model

• Disadvantages:– Modeling is usually limited to 1-2 loci– Sensitive to ascertainment bias

• Examples of sensitivity to ascertainment bias

– Sampling may include lost of “sporadic” families, not appropriate for linkage analysis

– What do “negative” results imply?– Confounders can include heterogeneity, multilocus

models

Familial clustering/recurrence risk to relatives

• Familial aggregation is the clustering of affected individuals within families.

• Methods for determining familial aggregation include:

• Family history approach (Khoury et al., 1993)

Family History Approach

Specifically

Ascertain the presence or absence of a family history of the disease in the study participants.

Three variations depending on the level of detail of information obtained on the relatives:

Abbreviated family history

Detailed family history

Family Study

Family History Approach 2x2

If you are performing a case-control study, you can treat family history as a “risk-factor” and use a standard 2x2 table to assess the significance

If table is

Odds ratio= (a*d)/(b*c)

a b

c d

+ -Family History

Disease in study participant

+

-

Correlation coefficients

• Quantitative traits can be plotted using a trait value in one relative versus another.

• The slope of the line is the square of the correlation coefficient.

• If no correlation, then environmental factors predominate.

R2 example• Specifically,• e.g.

0

10

20

30

40

50

60

70

80

40 45 50 55 60 65 70

height

Mean Parental Height

Mean Offspring height

If you have a quantitativetrait, you can polot the traitvalue in one relative versusthe other

The slope of the line is thesquare of the correlation coefficient.

This graph suggests that mean offspringheight is primarily a consequence ofhereditary factors.

Recurrence Risk to relatives

• A measure of how “genetic” a trait or disease is. What is the rate of affection for relative of proband with the disease versus the frequency of the disease in the in the general population? (Risch, N. Am. J. Hum. Genet. (1990): 46: 222-253.

• Sibling Recurrence Risk: S

iprelationsh of degree theindicates R'' where

population generalin rate

proband of relativein rate recurrenceR

Interpretation of S

• Values > 1.0 are generally taken to indicate evidence in favor of a genetic component. In general, the higher the value, the stronger the genetic component.

• Values can be used to estimate the number of genes under different genetic models.

• Note that the magnitude of the estimate is very dependent on the frequency in the population. For example, a common disorder may have frequency estimates of 3-6% depending on how a given study was performed but this results in small lambda.

Twin and adoption studies

• Twins are special cases of siblings– MZ twins share 100% of their genetic

material– DZ twins share 50% of their genetic

material (same as other sibs).

Twin Studies Continued

• Comparisons of Disease concordance rates between MZ and DZ twins provide information regarding the involvement of genetics

– Concordance rates with MZ>DZ are consistent with the involvement of genetics

First Twin Study Example Examples: One twin is affected, how often is

the other affected?

deviation from the expected frequencies may be due to incomplete penetrance

MZ DZType of Disease

90% 90%Probably environmental

100% 25%Mendelian recessive

Second Twin Example

MZ DZType of Disease

80% 16% ??

72% 35% ??

7% 7% *

*Especially may be the same as the population frequency;Therefore this may not be a good model for hereditary disease.

Twin Studies

• Twin studies are believed to control for many confounding environmental effects:– Age– Common familial environment

• Objection: Applicable for childhood, but not necessarily prenatal and rarely true for adult exposures

Cautions on Twin Studies

• Misclassification of zygosity

• Control for sex• Diseases with variable age of onset

Placenta Chorion Amnion DZ(%) MZ(%)

2 2 2 50 15

1 2 2 50 15

1 1 2 - 70

1 1 1 - rare

Adoption Studies

• Like twin studies, adoption studies can be used to test for evidence of genetic versus common familial environmental factors in the etiology of a disease.

• Cases are ascertained and the frequency of the disease in the biological parents of the case is compared with than in the adoptive parents.

• Difficulties include:– Achieving the necessary sample size– Similarity of adoptive and biological parental

environments

Adoptive Twin Approach Variation on Twin Studies, ‘Twins Reared Apart’. Should be very

sensitive at teasing out environmental vs. hereditary causes of disease.

Frequency of disease in biological parents

Frequency of disease in adoptive parents

Possible Etiology

85% 4%

Implies genetic etiology, frequency in adoptive parents may reflect general population risk

4% 85%Implies common environment is the primary risk factor

Heritability

• A statistical measure of the degree to which a trait is genetically determined

• Conceptual defininition: the ratio of genetic variance to total variance

• Rigorous definition: The proportion of the total phenotypic variance (V) of a trait that is due to genetic variance (G). This is heritability in the “broad sense”. It is a form of Principal Components Analysis.

Heritability• Approach: Compare variation among different

classes of relatives to that predicted by simple genetic factors

• Variance can also be broken down into additive, dominant, and epistatic variance.

• Correlation coefficient revisited: heritability= fraction of genes shared via identity by descent times the r squared. This represents heritability in the narrow sense (additive variance).

• Heritability can be more rigorously defined in twin studies:

• h2=2(rMZ2-rDZ

2)

Co-occurrence with other genetic conditions

• This occasionally occurs in a subset of patients and can provide very useful clues to the location of disease genes.– Examples:– Association between Trisomy 21 and Alzheimers

disease led to the isolation of the APP gene.– Association between chromosome 15q11-q13

duplications and inversions with autistic disorder provides one of the most promising candidate regions for an autistic disorder locus.

Experimental systems

• Animal Models– The occurrence of a human disease trait in

a model system can provide support for the genetic basis of human disease, and a method for investigating the genetic mechanisms.

• Example:– Over 40 mouse gene have been implicated in NTD

(Harris and Juriloff, Teratology 56:177-187 (1997)).

Inheritance mode consideration (with statistical

implications)

• Mendelian subsets?– Early onset?– More severe?– Clear-cut transmission allows parametric

transmission analysis?

Consider Study Design

• Affected sibling pair– Easy to collect?

• Affected relative pair/extended pedigree– Better for fine mapping?– Allows consideration of different genetic

models?

Family triads

• TDT or some variant?

• Assist with fine mapping?

• Include families not necessarily multiplex?

• Increase generality of results?

Case-control

• Easy to collect?

• Problems with establishing good control group?

The TDT

• As a test for linkage, can use singleton families, affected sib families, and extended families

• Looks only for informative familial relationships

• Has greater sensitivity than linkage analysis or affected sib pairs

A TDT example

nontransmitting

transmitting

1 x

1 78

x 46

•79 parents heterozygous for Class I allele at insulin gene polymorphism•(other alleles = X)•Triad results: 24 transmitted class I, 10 transmitted X•Affected Sib Pair Families:•15 transmitted class I to both children•24 transmitted class I to one child and X to the other•6 transmitted X to both children

004.0

26.84678

)4678( 2

valuep

TDT

Evidence of linkage with the TDTAllele-sharing test is not significantFor linkage (p-value=0.20)

Special populations

• Homogeneity?

• Generality?

Advantages and Disadvantages of the TDT

• Advantages– Can use singleton data– Can be more powerful than ASP tests

(even when the same data are used)

• Disadvantage– Has little power to detect linkage unless

there is also allelic association

Subject Recruitment

– Researchers responsibility to subjects• Maintaining IRB protocols ensuring adequate

review and oversight of project• Sharing results with subjects?• Maintaining contact through

– Newsletters?– Web site?– Phone access?

Sample Collection

• Blood for DNA

• Cell Lines

• Other tissue to allow extraction of tissue specific RNA– e.g., muscle biopsy for muscular dystrophy

or tumor tissue for cancer research, FMRP through scalp hairlet bulbs.

Molecular Design

• Genomic Screen

• Candidate gene analysis– Candidate-by-function?– Candidate-by-location?– SSCP to screen or sequencing for more

accurate gene assessment?

• Expression studies

Translation to Clinical Care

• DNA for presymptomatic and/or diagnostic testing?

• Develop prevention/treatment/cure?

• Phamacogenetics initiatives?