tcga glioblastoma pilot project: initial report

TCGA Glioblastoma Pilot Project: Initial Report

June 2008

NCI-NHGRI Partnership

Cancer Genomics: Ultimate Goal

Create comprehensive public catalog

of all genomic alterations present at significant frequency

for all major cancer types NCAB Report Feb 2005

Genomic Alterationscopy-number alterationstranslocationsgene expressioncoding mutations methylation

Integrate

NCI-NHGRI Partnership

TCGA Pilot Project Launched, 4Q2006

Cancers: • Glioblastoma multiforme• Squamous cell lung cancer• Ovarian cancer (serous cystadenomacarcinoma)

Goal: • Assemble high-quality samples of each type

• Characterize tumor genome by various approaches

• Rapidly share data with scientific community

• Compare and improve technologies

• Integrate and analyze data to illuminate genetic basis of cancers

Key questions posed at start of project

1. Can samples of adequate quality and quantity be assembled?

2. Can high-quality, high-throughput data be generated with current platforms?

3. How sensitive, specific and comparable are current platforms?

4. How can diverse data sets be integrated -- and what can be learned from integration?

5. Can recurrent events be distinguished from random background noise?

6. Can we identify new genes associated with cancer types?

7. Can we identify new subtypes of cancer?

8. Does new knowledge suggest therapeutic implications?

9. Can a network project drive technology progress in cancer?

TCGA Components

(1) >80% tumor cell content

(2) <40% necrosis

(3) Matched normal DNA sample

(4) Clinical annotation

(5) Appropriate informed consent

Current collection > 200 high quality tumors

GBM Samples: Strict Criteria

TCGA GBM: Center Overview

Broad/DFCI

Harvard LBNL MSKCC JHU/USC Stanford UNC SequencingBroad, WU, Baylor

SNP 6.0

Copy Number

HTA

RNA Expression

aCGH

Copy Number

Exon Array

RNAExpression

aCGH

Copy Number

GoldenGate

Methylation

Infinium

Copy Number

2 color arrays

RNA, miRNAExpression

PCR >ABI

SomaticMutations

Glioblastoma samples

Cross-platform Data Integration, Comparison

Cancer genomes are complex

Individual cancer genomes Integrate across many samples

New analytical methods needed

PresentationsCameron Brennan GBM and Genome Characterization

Stephen Baylin The GBM Epigenome

Rick Wilson Identifying mutations in GBM and

application of next gen sequencing technologies

Charles Perou The Challenge of Integrative Analysis

Eric Lander Summary and Discussion

Copy-number alterations: Discordance in initial studies

208 97

144

Event counting

n=141

n=37n=178

Little overlap in early studiesInclude only some known genes

Statistical significance

27

26 24

Many fewer eventsHighly concordantIncludes all known genes

AmplificationsDeletions

Copy-number alterations: ~30 in GBM

Coding mutations: Assessing statistical significance

Total across S samples

total bases

Random mutations: 2 x 10-6 per base 0.3% per typical coding region

Cancer-related mutations: Aim to detect 3-5% per typical coding region

Significantly mutated genes in GBM

600 genes x 86 GBM (non-hypermutated)

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

ProNeural Normal-like MesenchymalEGFR

Integrated analysis defines four subtypes in GBM

• Copy-number alteration

• RNA Expression

• DNA sequencing

• Methylation

EGFR ERBB2

PI-3KClass I

PI-3KClass I

PDGFRA MET

mutation, amplificationin 45%

mutationin 7%

amplificationIn 13%

amplificationin 4%

RASRASNF-1NF-1

AKTAKT

FOXOFOXO

PTENPTEN

ProliferationSurvival

Translation

Activated oncogenes

MDM4MDM4

TP53TP53

MDM2MDM2

CDKN2A(ARF)

CDKN2A(ARF)

RB1RB1

RTK/RAS/PI-3Ksignaling network

86%

RTK/RAS/PI-3Ksignaling network

86%

P53signaling

86%

P53signaling

86%

Senescence Apoptosis

CDK4CDK4

CDKN2A(P16/INK4A)

CDKN2A(P16/INK4A) CDKN2BCDKN2B CDKN2CCDKN2C

G1/S progression

homozygousdeletion in 51%

RBsignaling

77%

RBsignaling

77%

homozygousdeletion in 47%

homozygousdeletion in 2%

homozygousdeletion in 49%

amplification in 14%

amplification in 6%

mutation, homozygous

deletion in 35%

amplification in 17%

homozygous deletion,mutation in 11%

mutation in 2%

amplification in 2%

mutation in 2% mutationin 15%

mutation, homozygousdeletion in 17%

mutation, homozygousdeletion in 36%

CDK6CDK6CCND2CCND2 amplification in 1%

amplification in 2%

Pathway Analysis in GBM

454 Solexa, ABI SOLiD

~200bp

400 K / run

400 Mb

30+bp reads

~40 M / run

8-12 Gb

~700 bp

~100/run

70 K

ABI 3730XL

Next-generation sequencing technology

Read length Read number

Total bases

+

others

Costs decreasing rapidly

Key questions posed at start of project

1. Can samples of adequate quality and quantity be assembled?

2. Can high-quality, high-throughput data be generated with current platforms?

3. How sensitive, specific and comparable are current platforms?

4. How can diverse data sets be integrated -- and what can be learned from integration?

5. Can recurrent events be distinguished from random background noise?

6. Can we identify new genes associated with cancer types?

7. Can we identify new subtypes of cancer?

8. Does new knowledge suggest therapeutic implications?

9. Can a network project drive technology progress in cancer?