René Bernards
Nieuwe ontwikkelingen en hun vertaling naar
de kliniek
The Netherlands Cancer Institute
Amsterdam
Olympus microscope, late 1900sVan Leeuwenhoek microscope, late 1600s
The microscope, a major tool for cancer
diagnostics for the last 350 years
• Grade 3• Grade 1
Histological grade
Using microscopy to predict
disease outcome in cancer
Low risk High risk
Molecular diagnostics, better than the microscope?
Micro-scope Micro-array
MammaPrint: a 70 Gene Prognosis Profile
van´t Veer et al., Nature 415, p. 530-536, 2002
70 significant prognosis genes
Tum
or
sam
ple
s
High riskLow riskMammaPrint®
The molecular pathology challenge: how to integrate the old and the new?
High riskLow risk
Historical Grade versus MDx
Grade 3Grade 1
Low risk High risk
Grade 2
“Intermediate”
0%10%20%30%40%50%60%70%80%90%100%
Grade1
Grade2
Grade3
70-genehighrisk
70-genelowrisk
Responses to cancer drugs are
also determined by the genes!
Conventional cancer drugs are not very effective
One in four women benefit
from chemotherapy
Young women < 50 year Older women > 50 year
One in thirty women benefit
from chemotherapy
Breast cancer mortality
J Natl Cancer Inst. 2009;10:1044-8Targeted cancer drugs are very expensive
and often fail to yield clinical benefit
NSCLC
Breast
Pancreas
Renal
Strong correlations between cancer genotype
and drug response
• HER2 expression (breast) Trastuzumab
• BCR-ABL translocation (CML) Gleevec
• KRAS mutation (colon) Cetuximab (no response)
• EGFR mutation (lung) Erlotinib
• BRAF mutation (melanoma, colon) PLX4270
• ALK mutation/translocation (neuroblastoma, lung, breast) PF02341066 (crizotinib)
Personalized medicine 101
20 patients will validate your biomarker!
Some genotype - drug response
relationships are readily explained
Grb
RAF
EGFR
KRAS
MEK
ERK
Cetuximab
Wild type
Clinical response
Grb
BRAF
RTK
RAS
MEK
ERK
PLX4270
Clinical response when mutant
Downstream
pathway activation
Oncogene
addiction
Mutant
No clinical response
KRAS
Different types of breast cancer have different
mutations
Triple negative ductal
Carcinoma
Invasive lobular
CarcinomaPIK3CA
Mutation: 8% 40%
Response to
chemotherapy: good poor
Sorlie et al. (2001), PNAS 98, 10869
Hierarchical clustering of breast cancer gene
expression defines distinct molecular subtypes
BluePrint: Molecular subtyping Profile
Three classes were determined by
IHC in concordance with
TargetPrint microarray single gene
read-out:
– Luminal-type = positive ER and/or PR
hormone receptor status (hormone
positive)
– HER2-type = positive HER2 status
– Basal-type = negative ER, PR and HER2
status (triple negative)Dataset used: NEJM-295 Rosetta data
Profile: 80 genes (overlap with Perou ~25 genes, overlap
with PAM50 9 genes, overlap with MP 4 genes of 70-set, 10
genes of 231-set, overlap with TP 3 genes).
Development of BluePrint Molecular Subtyping Profile:
TargetPrint Single
Gene Expression
Concordance
IHC Testing
Basal-TypeHER2-TypeLuminal-Type
Measuring estrogen receptor function by
analysis of 80 downstream targets
•3 classes:–Hormone receptor positive
–HER2 positive
–Triple negative
Luminal reporter genes often have ER binding
sites in their promoters
32 out of 58 luminal signature genes
have an ERE in their promoter
p= 1.2 .10-5
Collaboration:
Wilbert Zwart
Jason Caroll
Cambridge UK
The response to chemotherapy varies in the different molecular subtypes
in-silico analysis of patients (n=133) described in Hess et al. 2006 JCO 24:4236
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
pCR
RD
pCR 50% 56% 9%
RD 50% 44% 91%
Her2+ Triple - Hormone +
Basal-type 50% pCR
HER2-type 56% pCR
Luminal-type 9% pCR
Basal and HER2-type show significant higher pCR rates
Molecular subtypes also exist in colon cancer
and have different mutation
Hierarchical Clustering training set (n= 188, whole genome)
- reveals 3 molecular subgroups with different outcome
High throughput “kinome” sequencing in breast
cancer
• 518 putative kinases
• 12 PI3K domain-containing genes
• 6 PI3K regulatory components
• 13 diglyceride kinases
• 18 genes frequently mutated in human cancer
• 19 genes specifically known to be mutated in breast cancer
586 genes total analyzed
SureSelect Exon capture technology: allows isolation of specific sequences from genome
Capture
HMEC
BT474
PB CF001
PB CF003
PB CF004
XG BC60P
XG MT201
XG HBCX17
Deep sequencing identifies regions of copy number
gain and loss
All regions shown on chromosome 17
ERBB2
Overview copy number alterations in breast tumor
samples
Erbb2 Erbb2
PTENCDKN2A/B (no coverage)
Each point represents the average of all regions for a particular gene
Gene level assessment can miss small structural changes
TAOK1, NEK8, NLK, KSR1
EXAMPLES OF SOME SNV found in COSMIC
HMEC BT474 XG BC60P XG MT201 XG HBCX17 CRC met TN001 TN002
APC - - - - - L1129S - -
EGFR - - - - - - E872Q -
KRAS - - - - - G13D - -
PIK3CA - K111N - - - - - -
TP53 C176P E285K R175H - - - - R248W
FRAMESHIFT/STOP CODON
TP53 BRCA2 GEN1 TP53
SPEG BRCA2
ACVR2B C9ORF96
AMPLIFIED (>3 fold)
MYC ERBB2 TTBK1 ERBB2 NRBP2
CRKRS PTK7 TAOK1 PTK2
PIP4K2B NEK8 MAPK15
RPS6KB1 NLK ADCK5
PDK2 KSR1 MYC
AURKA DGKE IKBKB
COL1A1
STK4
HOMOZYGOUS DELETIONS
PTEN JAK1 (E6) CDKN2A
BMPR1A CDKN2B
Summary SNV and CNA
Genotype-driven clinical trials
Conventional cancer treatment
Patient’s
tissue sample
pathology
grade, size, IHC
Chemotherapy
Personalized cancer treatment
Patient’s
tissue sample
Targeted (combination)
therapy
PPP
PP PPP
P P P P P P
PP
P
Molecular
diagnostics
which pathways
are active?
Conventional cancer treatment:
Rx
Treatment
Chemotherapy
Dx
Diagnosis
Stage, Grade, IHC
Personalized cancer treatment:
Rx
Treatment:
Pathway targeted therapy
Rapid progress in DNA sequencing
2000: first human genome:
cost $ 2,700,000,000
2010: many cancer genomes
cost: $ 10.000
Extrapolate to 2020: cost $ 0.04 per genome
Conclusion:
Need to change tissue preservation
Frozen
RNA integrityFFPE Protein integrity
The future of personalized medicine?
Acknowledgments
MammaPrint
Laura van ‘t Veer
Marc van de Vijver
Stephen Friend
Kinome sequencing
Ian Majewski
Astrid Bosma
NKI screening center
Wouter Nijkamp
Roderick Beijersbergen
Microarray facility
Ron Kerkhoven
Arno Velds
Our funding agencies:
Annuska Glas
Arno Floore
Iris Simon
Paul Roepman
Femke de Snoo