The Predictivity Concept

1
The Predictivity Concept

• Peter Propping
• Institute of Human Genetics
• University of Bonn, Germany

CDBI Seminar on predictivity, genetic tests and
insurance Strasbourg, 3-4 December 2007
2
Gene-environment Interaction
Source Dr. Ron Zimmern, Oxford
3
The Human Genome 3,2 x 109 nucleotide pairs
not a unique sequence, but appreciable
interindividual variation
any two genomes 99,9 DNA sequence
identity, thus, 0.1 sequence differences (3
mio). Any individual (diploid, i. e. two
genomes) 6 mio differences to the reference
genome.
4
Modes of inheritance
5
Two major groups of genetic diseases

• Monogenic ( Mendelian) disorders
• - monocausal
• clear relationship between genotype and
  phenotype
• - about 2.000 disorders clarified
• - most disorders are rare
• - therapy mostly difficult
• Genetically complex (multifactorial) disorders
• - complicated genetic structure
• - many of them common in the population
• may be influenced by exogenous factors
• therapy frequently possible

6
Conceptual distinction - Prognosis statement
about the future course of a past or currently
existing disorder - Prediction probability of
the onset of a disease that has not yet
occurred
7
Methods of prediction and prognosis in a
proband - medical history - medical
examinations - family history - predictive
genetic diagnosis - prediction based on lifestyle
8
Prediction on the basis of medical
examinations Imaging techniques (CT, MRT,
Ultrasound) - e.g. polycystic kidney
disease hereditary brain tumors, e.g. tuberous
sclerosis degenerative brain disorders Electroc
ardiogram - e.g. hereditary disturbance of
conductivity (long QT-syndrome) Blood
biochemistry - e.g. hypercholesterolemia hyperli
pidemia
9
Genetic diagnostics in familial adenomatous
polyposis (FAP)
10
Predictive diagnostics in familial adenomatous
polyposis (FAP)
11
Persons at risk for Lynch Syndrome (Hereditary
Nonpolyposis Colorectal Cancer, HNPCC)
12
Cumulative risk in carriers of amutation in the
BRCA1 or BRCA2 gene
Meta-analysis, King et al., Science 2003
13
Examples for Hereditary disorders with late onset
for which predictive genetic diagnosis is
possible (autosomal-dominant) Treatable Heredit
ary tumor syndromes - breast/ovarian
cancer - colorectal cancer - familial adenomatous
polyposis Polycystic kidney disease, type
1 Hereditary deafness, several late onset
forms Untreatable Huntington disease Myotonic
dystrophy Alzheimer disease, autosomal-dominant
forms Spinocerebellar ataxia, several
forms Facio-scapulo-humeral muscular
dystrophy Retinitis pigmentosa, several late
onset forms
14
Concordance rates in identical (monozygotic, MZ)
and fraternal (dizygotic, DZ) twins
MZ DZ Coronary heart disease
46 12 Hyperthyroidism 47 7 Neurodermitis 83
28 Diabetes mellitus I 45 5 Diabetes mellitus
II 95 10 Lepra 59 20 Epilepsy
(idiopathic) 86 4 Schizophrenia narrow
definition 26 4-10
wide definition 41 10-20
15
Genetic model of a complex (multifactorial)
disease Hypertension as an example
super-normal
slightly predisposed
slightly increased
definitely increased
severely ill
16
Relationship between genotype and phenotype in a
complex disease - Predictive value of a
genotype Positive Predictive Value (PPV)
- Fraction of persons with a predisposing
genotype who will develop the disease
Negative Predictive Value (NPV) - Fraction of
persons without the genotype who do not
have the disease
17
Positive Predictive Value (PPV) Example Crohn
disease and association with NOD2 variant
Positive predictive value Homozygous 20
0,95 Heterozygous 57 0,71
20 21
57 80
18
Relationship between Genotype Frequency, Relative
Risk and Positive Predictive Value
Disease Disease Genotype Genotype Relative
Risk PPV Frequency Frequency COPD 0,05 Pi
ZZ 0,0005 20,0 99,1 Narcolepsy 0,0005 DQB1060
2 0,021 10,5 0,4 homozygosity COPD
chronic obstructive pulmonary disease
19
Predictability of affection status in the carrier
of a predisposing genotype - monogenic
diseases up to 100 depending on penetrance -
complex (multifactorial) diseases often low
eventually higher after genotypic profiling
20
To what degree can multifactorial disorders be
predicted ?
Generally, the concordance rate of MZ twins is
the upper limit of prediction but only
cross-sectional information taken into account,
no age correction possible global concordance
rates give only average data, in fact part of the
cases higher degrees of heritability may exist.
21
Screening approaches - Genetic population
screening newborn screening for treatable
diseases e. g. preconceptual thalassemia
screening on Sardinia and Cyprus
preconceptual screening in certain ethnic groups,
e. g. for Tay-Sachs in Jews cascade
screening, e. g. for hypercholesterolemia in the
Netherlands
22
Screening approaches - Ascertainment of persons
at high risk through family history e. g.
inherited breast/ovary cancer and Lynch syndrome
(HNPCC) population-based for
preconceptual testing in recessive diseases
23
The Future The 1000 Dollar Genome
- nightmare of informed consent - nightmare of
interpretation