Transcript Epidemiology in the postgenome era – towards a less reductionist

Biobanks of
Cerice
Center for Gene Expression Research in Cancer
Epidemiology
Eiliv Lund, UiTø
• How should the description of the
genome change epidemiological design
if this was to be a change of paradigm
for medical research?
Functional genomics
• Two major components
• Determining the sequence of the genome
• Patterns of gene expression
Cohort studies
Information content in cohort studies
- questionnaire (interview) information
- biomarkers (plasma, serum, urin etc) –
proteom/metabolom
- DNA for SNPs analysis (buffy coat, full
blood)
- Eventually parafin blocks for tumor DNA
Present situation for the
traditional cohort design
• Huge amount of cohort fullfilling most of
the criteria
• EPIC – European Prospective Investigation
into Nutrition and Cancer
• Consortium of cohorts (NCI)
• New cohorts – Estonian, Biobank UK, the
Last cohort etc
The Norwegian Women and
Cancer postgenome cohort study
• The intention was to add two new biobanks:
• 1. A prospective collection of blood for
expression analysis, whole genome
microarray analysis
• 2. Tumor samples from breast cancer cases
arising in the cohort preserved for
expression analysis
Questionnaires
Passive Follow-up – Cancer Registry
NOWAC all women,
n=170 000
NOWAC
Biopsy cohort
Women born 1943-57
n=147 000
Active
follow-up
through
biopsy
collection
NOWAC
Expression
cohort
n = 40 000
5 years
30 Years
3 new designs
• How to integrate expression analysis into
standard prospective design
• Why
1. Prospective expression
analysis
• Fingerprints of exposure
• Metabolic pathways
• Early diagnostic markers
# genes=16164
HRT vs. no HRT
Cod-liver oil + capsels vs. no cod-liver oil
Cod-liver oil vs. no Cod-liver oil
Cod-liver oil in capsels vs. no cod-liver oil
2 sample t-test Significant genes, BAM
p<0.01
61
3
622
219
243
0
417
5
1. Prospective expression
analysis
• Fingerprints of exposure
• Metabolic pathways
• Early diagnostic markers
2. Case-control upon a cohort
design
• Information collected as a case-control
study at time of diagnosis AND the same
information at start of follow-up
• Intra-individual gene expression analysis
• Repeated information
3. Prospective changes of gene
expression in peripheral blood cells in
relation to expression profiles of breast
cancer tissue
• Case-case design for breast cancer cases
comparing expression patterns over time in
relation to tumour expression, eventually
stratified on genotype, adjusting for lifestyle
exposure like hormones etc.
Biopsy project
• Inclusion criteria born 1943-57
• Collaboration with the nationwide research group
Norwegian Breast Cancer Group, NBCG
• Breast cancer only treated at public hospitals
• Excluding hospitals with less than 20 cases per
year, or 9% of total case load. Increasing
centralisation due to mammographic screening
programme leaves us with about 20 hospitals
Statistical power
• Expression cohort: 40 000 women gives
approximately 120 breast cancer cases each year
(incidence rate 300/100 000 per year). Inclusion
rate unknown – estimated to 60% – follow-up time
5 years, funded by NFR 2005-07
• Biopsy cohort: 150 000 women gives
approximately 450 breast cancer cases per year
Design and statistical power
• Change from indirect to direct design
reduces the sample size till ¼ - ½
Conclusion
• The NOWAC postgenome cohort need to add new
designs for incorporation of expression analysis of
peripheral blood and tumour tissue in standard
prospective studies
• The collection of the first biobank of buffered
RNA for prospectively use will be completed
winter 2006
• The adding of a tumour tissue bank for breast
cancer will start winter 2006
•
Thanks to A-L Børresen Dale for enthusiastic collaboration