Cancer genome sequencing - PowerPoint Presentation

Slide1

Cancer genome sequencing

Tim Graubert, MD

Division of Oncology, Stem Cell Biology Section

Washington University School of Medicine

Siteman Cancer Center

Slide2

Genome Center at Washington University

Slide3

Genome Center Leadership

Rick Wilson

Elaine Mardis

Tim Ley

Slide4

The Cancer Genome Atlas Project

is studying the genomes of adult cancer patients and their tumors to identify genetic changes that underlie cancer.

1,000 Genomes Project

seeks to catalog the immense human variation written into the genetic code.

Washington University Cancer Genome Project

aims to sequence the tumor and matched normal genomes of hundreds of cancer patients.

Pediatric Cancer Genome Project

is a collaboration with St. Jude Children’s Research Hospital to identify the genetic changes that give rise to some of the world’s deadliest childhood cancers.

The Human Microbiome Project

is utilizing genome technology to catalog the species of bacteria that reside within human body is states of health and disease.

Current major projects

Slide5

One exon at a time, one gene at

time

Very slow, expensive, and

cumbersome

Yield thus far has been limited: we don’t know where to look

Discovering cancer-causing mutations

with biased screens

Slide6

Cancer genetics:

how do we find what we don’t know?

Candidate gene (“list-based”) sequencing

Unbiased screens

methylome

transcriptome

genome

Slide7

What kinds of changes need to be detected?

Both inherited and acquired (somatic)

Big ones

translocations

, inversions

copy

number

alterations

Little ones

single nucleotide variants

small

insertions and deletions

Epigenetic ones

expression changes

methylation

changes

chromatin

modification

Slide8

whole genome sequencing

bone marrow

aspirate

tumor

DNA

WGS

~30X

SNVs

Indels

SVs

skin

biopsy

normal

DNA

WGS

~30X

SNPs

Indels

CNVs

somatic

Tim Ley

Rick Wilson

Elaine Mardis

Slide9

Tumor and skin

samples

2

paired-end libraries (200-250 and 350-400

bp

fragments

)

30x haploid coverage (~99% diploid)

Variant calling (SNVs,

indels

, structural)

Remove variants found in skin (inherited) and other genomesPrioritize variants for validation (Tiers)

Validate variants with PCR (custom capture) and digital read countsAssess mutation frequency in other samplesdeposit in dbGAP

Functional validation (long term)

analysis pipeline

Slide10

Prioritizing potential mutations into non-overlapping tiers

48.7%

1.3%

8.6%

41.4%

Tier 1:

genic

Tier 2: conserved/regulatory

Tier 3: unique

Tier 4: the rest

Slide11

Caucasian female, mid-50s at diagnosis

De novo

M1 AML

100% blasts in initial BM sample

Relapsed and died at 21 months

Normal cytogenetics

No LOH/CNV on Affy 6.0 or Illumina 1M SNP arrays

Informed consent for whole genome sequencing

Ley et al.,

Nature

2008

Acute Myeloid Leukemia (AML)

Slide12

0

20

40

60

80

100

% Variant

CDC42

S30L

NRAS

G12D

IDH1

R132C

IMPG2

G834D

ANKRD26

K1300N

LTA4H

F107S

FREM2

Q2077E

C19orf62

e5-1

NPM1

c

rs60702183

rs9636146

rs12358887

Primary Tumor

Skin

rs7396397

rs6966150

rs2960642

CEP170

177insL

SNPs

Mardis et al NEJM 2009

AML2 (M1): 86% Blasts

Slide13

IDH1 mutations in AML

IDH1

encodes for isocitrate dehydrogenase.

IDH1

R132

mutations identified in 16 of 188 (8.5%)

de novo

AML patients.

IDH1

mutations in AML associated with normal cytogenetics and poor prognosis.

Munich Leukemia Laboratory: IDH1 mutations in 9.3% of AML (n=999)

strongly associated with an unfavorable prognosis.

Overall Survival (months)

Survival Probability

No IDH1 mutation (n = 172)

IDH1 mutation (n = 16)

Mardis, et al

NEJM 2009

Slide14

AML2

Tim Ley

Slide15

tAML1

Dan Link

Slide16

How many genomes to find the common recurrent mutations?

Assume cytogenetically normal “diploid” genomes:

30-40 genomes need to be sequenced to find 95% of the mutations that occur at a frequency of 5%.

WashU

Cancer Genome Initiative:

38 cytogenetically normal

(2 M0, 12 M1, 11 M2, 7 M4, 6 M5)

12 M3 with t(15;17) only

Slide17

FAB subtype

Cytogenetics

FAB subtype

MO

M1

M2

M3

M4

M5

Cytogenetics

Normal

t(15;17)

AML1-46

Tumor

Normal

2.5 Tbp: AML

2.40 Tbp: skin

Tim Ley

AML50 progress: 12/01/09

Slide18

AML50 progress: 2/16/10

6 Tbp: AML

5.7 Tbp: Skin

Tim Ley

Slide19

TCGA AML Project

500 AML cases

All subtypes

Multiple platforms:

Exome (Broad)

Copy Number (SNP array)

Transcriptome (Marra)

Epigenome (Laird)

Slide20

WU Cancer Genome Initiative

Genome sequencing of 150 cancer patients

(tumor & normal genomes)

50 AML cases

50 breast cancer cases

30 lung adenocarcinoma cases

12 glioblastoma cases (TCGA)

5 ovarian carcinoma cases (TCGA)

Others: prostate, pancreatic, multiple myeloma

12-month time line (Initiated 4/1/09)

Develop methods, pipeline and analysis tools.

Slide21

WU Cancer Genome Initiative

2 March 2010

WU Cancer Genome Initiative

Slide22

African-American female, mid-40s at diagnosis

Basal subtype (“triple negative”) breast cancer

Metastatic brain tumor (frontal lobe)

BRCA1/2 genotypes unknown

Deceased

Four samples:

PBL (normal)

Primary tumor

Metastatic tumor

Xenograft (“HIM”) of primary tumor

Breast cancer “quartet”

Matt Ellis

Slide23

Clonal evolution in breast cancer

Matt Ellis

Slide24

Pediatric cancer genome project

A collaboration between

The Genome Center

and

St. Jude Children’s Research Hospital

.

Initiated February 1, 2010

Complete genome sequencing of 600 pediatric cancer patients (in 3 years).

Leukemia (infant ALL, HR T-ALL, CBF AML)

Brain tumors (medullablastoma)

Solid tumors (NB, RB, osteosarcoma)First six cases completed.

Jim Downing

Slide25

Rapidly decreasing costs

Paired end read length

Gb

/run average

# runs/genome (30X)

*Cost per genome (30X)

Timeframe

35 bp

4Gb

25

$500K

Summer ‘08

75 bp

20Gb

10

$120K

Summer ‘09

100

bp

50Gb

2

$44K

January ‘10

100

bp

200Gb

1/2

$15K

June ‘10

*fully loaded direct cost, incl. validation

E Mardis

Slide26

Technical challenges of Prostate Genome Sequencing

Focal tumor type: low overall neoplastic cellularity requires LCM

FFPE preservation common

DNA isolated from tumor must satisfy library construction input

and

genotyping assay input

E Mardis

Slide27

Preserving library complexity with small samples

Adequate diploid coverage from <100 ng template.

Genomic DNA

1

μg

PE Library

100

ng

PE Library

10

ng

PE Library

Called SNPs

1,140,179

1,121,425

1,132,536

1,075,836

Het SNPs

329,645

325,705

325,567

269,332

E Mardis

Slide28

Conclusions and caveats

Next gen sequencing with paired end reads can identify all classes of inherited and acquired mutations in cancer genomes

False positive rate are falling, but all mutations still require validation

Hundreds of sequence variants per genome, but only a tiny fraction are probably relevant

The greatest challenge going forward: finding the important mutations

Slide29

The future challenges

Transitioning from one-at-a-time analysis to multi-case analysis

Integrating data sets from RNA-seq, methylation, miRNA-seq, etc. to formulate a more complete picture of somatic alterations

Cataloging germline variation to establish a framework for cancer susceptibility

Slide30

“If the goal is to solve the pathogenesis of cancer, there will never be a substitute for understanding the structure and sequence of the entire genome.”

- Renato Dulbecco, 1986

Slide31

Wash U Colleagues

Timothy J. Ley, M.D

.

John DiPersio, M.D.

Mark Watson, M.D.

Matthew Ellis,

M.B., Ph.D

.

Ramaswamy Govindan, M.D.

Peter Westervelt, M.D., Ph.D.

Jackie Payton, M.D., Ph.D.

David Gutmann, M.D.

Adam Kibel, M.D.

St. Jude Colleagues

James Downing, M.D

.

William Evans, Pharm.D.

Michael Kastan, M.D.

The Genome Center at WU

Rick Wilson, Ph.D.

Elaine Mardis, Ph.D.

Li Ding, Ph.D.

Ken Chen, Ph.D.

David Larson, Ph.D.

Michael McLellan

Daniel Koboldt

Christopher Harris

Lucinda Fulton

Robert Fulton

David Dooling, Ph.D.

Vincent Magrini, Ph.D.

a

nd many others

…

acknowledgments