The Goal

1. Understanding the etiology of disease X requires a genetic diathesis interacting with environmental, epigenetic and stochastic components.

Hasler et al., (2005)

The Problem

1. Understanding the etiology of disease X requires a genetic diathesis interacting with environmental, epigenetic and stochastic components.

2. Underlying genetic diatheses and environmental, epigenetic and stochastic mechanisms have remained mostly uncharacterized.

Hasler et al., (2005)

Source(s) of the Problem

1. Heterogeneity of phenotype (DSM Category)2. Diagnostic errors3. Environmental effects4. Multiple genes with small effects5. Gene X Gene interactions6. Gene X Environment interactions7. Epigenetic factors

How Bad Is The Problem?

p (disease|gene)• Odds ratio =

p (disease|no-gene)

Odd’s Ratio

Huntington's Smoking/Cancer

Psychiatric0

20

40

60

80

100

120

Kendler, 2005

Huntington’s

X Y

One-to-one relationship

Anxiety

•A V

•B W

•X Y

•C Z Many-to-many relationship

Types of Markers (Need not be Biological)

• Episode – associated only with active symptoms– May or may not be genetic– Not a genetic marker – would miss compensated or ‘at-risk’– May predict treatment outcome in specific subgroup, etc.

• Vulnerability – prone to illness, but not part of pathological genotype (black skin & sickle cell)

• Genetic – associated with the pathological gene– Linkage – non-allelic genes in close proximity are linked to disorder– Direct manifestation of genetic diasthesis

• These are endophenotypes

Iacono, 1988

Anxiety

•A V

•B W

•X Y

•C Z Many-to-many relationship

Anxiety

•A V

•B W

•X Y

•C Z Many-to-many relationship

Endophenotypes ?Fisher

(DysReg)

Associated with disease Yes

Reliable

Genetic

Identify at-risk

State independent

First-degree relatives

Plausible

Segregates with illness

Specificity

Endophenotypes ?Fisher Vasey

(Dysreg) (RSA)

Associated with disease Yes Yes

Reliable Yes

Genetic Yes

Identify at-risk

State independent

First-degree relatives

Plausible

Segregates with illness

Specificity No

Endophenotypes ?Fisher Vasey Hajcak

(DysReg) (RSA) (ERN)

Associated with disease Yes Yes Yes

Reliable Yes Yes

Genetic Yes Yes

Identify at-risk Yes

State independent Yes

First-degree relatives Yes

Plausible Yes

Segregates with illness

Specificity No No

Be Careful What You Ask For

• Flint & Munafo, 2007– COMT polymorphism & schizophrenia• COMT regulates dopamine

– Odds ratio = 1.13

• How about endophenotypes?– Move closer to the genes– Genes should account for more variance– Neuropsychological dysfunction associated with

prefrontal cortex

This Is What You May Get

• Wisconsin Card Sorting Task– Recognized measure of prefrontal function– Effect size of association = 0.5% of variance

This Is What You May Get

• Wisconsin Card Sorting Task– Recognized measure of prefrontal function– Effect size of association = 0.5% of variance

• N-Back task– Effect size was the same (0.5% of variance)

This Is What You May Get

• Wisconsin Card Sorting Task– Recognized measure of prefrontal function– Effect size of association = 0.5% of variance

• N-Back task– Effect size was the same (0.5% of variance)

• P300– Effect size was even worse (0.01% of the variance)

Schizophrenia

•A V

•B W

•X Y

•C Z

Conclusions• Odds ratios in psychiatric disorders are very low• Endophenotype strategy is very hot & plausible

-- at least theoretically

• Identifying a good endophenotype is difficult– Pay attention to the criteria– Don’t drop bomblets – do the work

• Current status– Some say only success to date is in schizophrenia– But Flint & Munafo is discomfiting

• Have fun – but “Be careful out there”.