Angel CarracedoFundación Gallega de Medicina Genómica

(SERGAS)

CeGen-ISCIII Universidad de Santiago

Estudio de asociación genética en la leucemia linfocítica crónica

XXXV REUNION DE LA ASOCIACION GALLEGA DE

HEMATOLOGIA Y HEMOTERAPIA. FERROL, MARZO 2011

Biological Complex Systems

• Why is it so complicated?

• Can we make sense of this complexity?

• Can we convey our understanding of this complexity?

Geness

Proteome

microRNA

Epigenetics

Environment

Systems

biology

Transcriptome

Genotype

Phenotype

Enviroment

Genes40%

Environment60%

CONTENT

- How to look for the genetic component of a disease

-An example with CLL

Better understanding of mendelian andcomplex disease

Better classification of diseases

Risk stratification

Pharmacogenetics and pharmacogenomics

Allelic heterogeneity

Locus heterogeneity

Phenocopy

Phenotypic variability

Trait heterogeneity

Gene-gene interactions

Gene-environment

interactions

How to look for low penetrance genes?

Genetic Strategies

Traditional (from the 1980s or earlier)

– Linkage analysis on pedigrees

– Allele-sharing methods: candidate genes, genome screen

– Association studies: candidate genes

– Animal models: identifying candidate genes

Newer (from the 1990s)

– Focus on special populations Haplotype-sharing (Jesus M. Hernández-Fam. Gam.)

– Congenic/consomic lines in mice (new for complex traits)-Animal models

– Single-nucleotide polymorphism (SNPs)-Whole genome scans (Association studies)

– Admixture mapping

– Functional analyses: finding candidate genes

mag

nit

ud

e o

f ef

fect

frequency of trait in the population

Linkage analysis of families

association studies in populations

obtainable sample size

Linkage analysis or association studies ?

•linkage analysis is usually more robust in the identification

of mendelian traits

• association studies have more power to detect genes with

small effects (Risch & Merikangas, Science 1996)

Phenotype A Phenotype B

Allele 1 Allele 2

SNP A is associated with Phenotype

SNP A:

Allele 1 =

Allele 2 =

Human Genetic Association Study Design

• 1,000,000 SNPs

SNP: SINGLE NUCLEOTIDE POLYMORPHISM

ATCGGCGTACCTGATTCCGAATCCGTATCG

ATCGGCGTACCTGAATCCGAATCCGTATCG

1000 PERSONAS -1,000,000 SNPS = 1,000 MILLONES DE ANÁLISIS Y DE DATOS

Characteristics of SNP Variation

• Clustering is observed on

all the autosomes:Haplotype blocks: Blocks

with little evidence

of recombination

• Some clusters appear

functional : MHC on

chromosome 6 (with

extensive replication)

Gabriel et al. Science, 296,2002

cM

Mb

1Mb windows

recombination hotspots

LD blocks (little or no recombination)

HapMap (2002)• Catalogue of variation at Single

nucleotide polymorphisms (SNPs) genome-wide in different populations

• Touted for disease gene identification via linkage disequilibrium mapping

• ‘Tag’ SNPs can cover whole genome

• Reduction of SNPs required to examine the entire genome for association with a phenotype from 20 million to 1,000,000 tagSNPs

• 1,000,000 SNPs

SNP: SINGLE NUCLEOTIDE POLYMORPHISM

ATCGGCGTACCTGATTCCGAATCCGTATCG

ATCGGCGTACCTGAATCCGAATCCGTATCG

1000 PERSONAS -1,000,000 SNPS = 1,000 MILLONES DE ANÁLISIS Y DE DATOS

Coordination

NODE 1Barcelona

(CRG)

NODE 2Santiago de

Compostela (USC)

NODE 3Madrid(CNIO)

Scientific International Committee Ethical International Committeel

Illumina Sequenom /

Affymetrix

Illumina

Spanish National Genotyping Center GeGen-ISCIII

0

5

10

15

20

25

30

CANCER

PSYCHIATRY

NEUROLOGY

ENDOC-METAB

RHEUMATOL

OPHTAL

CARDIOVAS

OTHERS

2005: 55 PROJECTS

2006: 75 PROJECTS

2007: 114 PROJECTS

2008: 135 PROJECTS

ASSOCIATION STUDIES CARRIED OUT IN CEGEN

2009: 4 GWAs2010: 15 GWAS

Association studies

Candidate gene approach

-Causative hypothesis or

candidate genes

Genome wide analysis (GWAs)

-No need of gene selection

-Lack of bias towards specific

genes

Both approaches are complementary

OXALIPLATIN

Previous case-control (association) studies to

identify common, low-penetrance cancer

genes

• Many small-scale studies in past, candidate

genes

• Many positive reports

• A priori p(false+) >>> p(true+)

• Publication bias, failure to match cases and

controls/population stratification, lack of

correction for multiple comparisons, lack of

replicationCorrection for multple comparisons

P> 10-7 required

TYPE I ERRORS: Population stratification

EPICOLON GWASPCA analysis on genotypes: checked genotyping dates, geographical origin, Nsp-Sty and collection hospital

Meixoeiro

Donosti

0.05

N=366

N=167

N=944

Type I errors: random

• Permutations (the most commonly used method-

computational intensive!)

• Other methods:

-False discovery rate (FDR)

-Sum Statistics

-Single Nucleotide Polymorphism Spectral

Decomposition

-Others

Corrections for multiple comparisons (p= 0.01

1 false positive every 100 comparisons)

• Bonferroni method

Pcor = 1-(1-Pnoncor)n new signif = alfa/n.

comparisons

-Very conservative-Assumption of independence

T. Manolio/ N Engl J Med 2010;363:166-76

Whole-genome association analysis

1 million

Genome-wide association study

(GWAS) to identify low-penetrance

genes

• Require many (>1000) cases and controls (but not always)-Consortia

• Can improve power by selecting cases (early-onset, familial) and controls (cancer-free)

• Search for alleles or genotypes over-represented in cases

• Verify in other sample sets

Chronic lymphocytic leukemia accounts for ~25% of all

leukemia and is the most common form of lymphoid

malignancy in Western countries.

Despite a strong familial basis to CLL, with risks in first-

degree relatives of cases being increased ~8-fold, to date the

inherited genetic basis of the disease is largely unknown.

.

Chronic lymphocytic leukemia

All association studies with candidate genes inconsistent

CLL

299,983 tagging SNPs

Stage 1: 505 cases and 1,438 controls (UK/Spain)

Stage 2: 180 SNPs in 540 UK cases

Stage 3: 19 SNPs

UK replication series 2 (660 cases, 809 controls)

Spanish replication series (424 cases, 450 controls).

Stage 4, 10 SNPs with the strongest association from a

combined analysis of stages 1–3 in a Swedish replication

series (395 cases, 397 controls)

T. Manolio/ N Engl J Med 2010;363:166-76

D. Crowther-Swanepoel D, Ana Vega, K.. Smedby, C. Ruiz-Ponte, J. Jurlander,

E. Campo, A. Carracedo, R. Houlston, British Journal of Haematology, 2010

Cumulative impact of 10 common genetic variants on colorectal cancer risk in 42,333 individuals from eight populations (Lancet, in press)

This study demonstrates that population subgroups can be identified with a predicted absolute CRC risk sufficiently high as to merit surveillance/intervention, although individualized CRC risk profiling is not currently feasible. Nonetheless, the findings provide the first tangible evidence of public health relevance for data from genome-wide studies in CRC

Spanish data GWAS

Birdsuite uses two different approaches for CNV detection:

- Canary: 1500 probes directed to CN Polymorphisms (as

described in the Human Variation Database browser)

-Birdseye: CNV detection

These data were also analysed with CNVAssoc

Preliminary results pending on stratification correction

From tagSNP to causal variation …..

Why is this important?

• Population portability

• Targeted interventions

• Learn more about how cancer develops

• Plan: Resequencing

• Check information from WGS

Nature last week: Identified the first CLL genetic mutations through NGS

NGS: SOLiD 4 System

Throughput: Up to 100 Gb/run

Fragment length:

Fragment: 50 bp

Mate-pair: 2 x 50 bp

Paired-end: 50 x 25 bp

Multiplexing:

96 DNA barcodes

48 RNA barcodes

Targeted resequencing

Exome sequencing

Whole genome sequencing

Panelso All Exon Kit (50 Mb Exome)oAll Exon Kit (38

o Mb Exome) (tiling 1x)o All Exon Plus Kit (38 Mb Exome + 3,3 Mb custom)

Customo < 200 Kbo 200 -500 Kbo 500 Kb – 1,5 Mbo 1,5 – 3 Mbo>3 Mb

Ion Torrent Personal Genome Machine (PGMTM)

Throughput: Up to 10/100 Mb/run

2012 - 1 Gb/run

2 hours/run

Fragment length:

Fragment: 100-150 bp

Unidirectional sequencing

Bidirectional sequencing

Multiplexing:

2011 - 96 DNA barcodes

SINGLE DNA MOLECULE SEQUENCING

Genetic variegation

of clonal

architecture

and propagating

cells in leukaemia,

Anderson et al.

Nature 2010

Isidro Sánchez-GarcÍa.

U. Salamanca

GWAS in pharmacogenetics-Differences with common diseases

Sample size: For ADRs the number of cases and controls can be much lower than for common diseases. However a number of published GWAs on pharmacogenomics have failed to show a large enough effect for genome-wide signifcance; the main reason for this is probably the small sample size with insufficient power to detect small or moderate effects.

Reasons: Phenotypic characterization- Some pharmacogenomics effects tend to be larger and involve fewer genes than in studies on common complex diseases.

Obtaining adequate number of cases for pharmacogenomics GWAs is more challenging than for common diseases. In many case serious ADRs often only affect on in every 10,000 to 100,000 patients treated.

Manhattan plot of −log P-value against chromosomal position of each marker from a study on simvastatin-induced muscle toxicity on 85 cases and 90 drug-exposed controls (A. Daly, 2009).

GWAS for pharmacogenomics

SNPs (chromosomal locations) shown previously to be associated with CRC risk are: rs6983267 (chr 8q24), rs4779584 (chr 15q23), rs4939827 (chr 18q21), rs3802842(chr 11q23), rs10795668 (chr 10p14), rs16892766 (chr 8q23), rs4444235(chr 14q22), rs9929218 (chr 16q22), rs10411210 (chr 19q13), rs961253 (chr 20p12).

5FU, oxaliplatino e irinotecan

Eficacia: RECISTToxicidad: CTC

Strong: Al menos un “grado 3-4” entre todos los efectos secundarios.Weak: Al menos un “grado 3-4” entre diarrea y náuseas, o al menos un “grado 1-2” en el resto de efectos secundarios (que se consideran más graves que diarrea o náuseas).Digestive: Al menos un “grado 3-4” entre diarrea y náuseas, o al menos un “grado 1-2” en mucositis.Circulatory: Al menos un “grado 1-2” entre leucopenia, trombopenia, anemia, y neutropenia.Others: Al menos un “grado 1-2” entre neuropatía y síndrome mano/pie.

EPICOLON GWAS (300 cases) 9 SNPs (p< 10-10) being replicated

OXFORD GWAS (620 cases, Capecitabine (5FU) and then randomised to oxaliplatin or no oxaliplatin, phenotypes by syntoms (i.e.Diarrhoea and Handfoot syndrome) 11 SNPs being replicated

EPICOLON GWAS TOXICITY

Methotrexate consolidation treatment according to pharmacogenetics of

MTHFR ameliorates event-free survival in childhood acute lymphoblastic

leukemia Running Title Methotrexate pharmacogenetics in childhood

acute lymphoblastic leukemia

Salazar et al. 2011 (Pharmacogenomics Journal ,submitted)

We investigated the usefulness of the MTHFR genotype to increase the

methotrexate dosage in the consolidation phase in 141 childhood ALL patients

enrolled in the ALL/SHOP-2005 protocol. Patients with a favourable MTHFR

genotype (normal enzymatic activity) treated with methotrexate doses of 5 g/m2

had a significantly lower-risk of suffering an event than patients with an

unfavourable MTHFR genotype (reduced enzymatic activity) that were treated

with the classical methotrexate dose of 3 g/m2 (p=0.012). Our results indicate

that analysis of the MTHFR genotype is a useful tool to optimize methotrexate

therapy in childhood patients with ALL.

Fenotipo Genotipo

Cuanto mejor definido es el fenotipo más fácil es encontrar el gen

del que depende. Cuanto más complejo es el sistema más difícil es

de definir su fenotipo

En todo aquello que tiene variación y esta es relevante clínicamente

se puede buscar el gen causal y así empezar a entender el fenotipo

y en consecuencia la enfermedad.