Standardization within the consortium

Cancer consortiaPaolo Boffetta

IARC

IARC-coordinated cancer consortia

• INTERLYMPH– >20 case-control studies of

lymphoma• ILCCO

– >40 case-control and cohort studies of lung cancer

• INHANCE– ~20 case-control studies of head and

neck cancer

Characteristics of IARC cancer consortia

• Emphasis on pooling of independently collected results

• Coordinated generation of new data• Projects proposed and managed by

working groups• Light central coordination• Expansion to low- and medium-

resource countries

Data flow

• No central facility for data management and analysis

• A common database has accumulated starting with the initial pooled analyses and including more and more data

• Contacts between people involved in subsequent analyses

Phenotype standardizationExample of InterLymph

• Pathological and genetic heterogeneity• Background

– reviews conducted within studies– need for network-wide review?

• Pathology working group– epidemiology-oriented classification

• hierarchical

– limited review (5• % of 10,000 cases)

Data flow - Steps

• Collection of study protocols, questionnaires and other forms– posted on website

• Data provided by PI– specific vs. free data format

• Checking and cleaning of data• Pooled analysis

– performed by working group– detailed preliminary results circulated among PI– test of heterogeneity among studies, sensitivity

analyses

Pooled AnalysesHead & Neck Cancer SNP

• 11 SNPs in metabolic genes, 7 in DNA repair genes • 10 case-control studies from the US and Europe • Request % undetermined and % concordance for

quality controls from each study• Test for heterogeneity by:

– Laboratory sources: genotyping method, source of DNA – Study characteristics: hospital vs. population-based, study

period, sample size – Other: ethnicity, age, smoking, alcohol drinking, subsite

• Standardization – adjust for variables that contribute to heterogeneity, present overall OR and stratified OR

Novel analysesImmunological SNP in

InterLymph• Selection of a list of relevant SNP

– 12 SNP related to immunological response

• Analysis of DNA samples in five laboratories– four used Taqman– one used Pyrosequencing or allele-specific PCR

• Quality control– 102 DNA samples from ethnically diverse

individuals that previously had been sequenced and genotyped (SNP500Cancer project)

– assays not in Hardy Weinberg Equilibrium (HWE) among controls were re-checked

Rothman et al., submitted