Colorectal cancer intrinsic subtypes are associated with prognosis,

chemotherapy response, deficient mismatch repair and epithelial to mesenchymal transition (EMT)

Josep Tabernero, Vall d’Hebron Hospital

and Iris Simon, Paul Roepman, Andreas Schlicker, Ian Majewski, Victor Moreno, Christine Chresta, Robert Rosenberg, Ulrich Nitsche, Teresa Macarulla, Gabriel

Capella, Ramon Salazar, George Orphanides, Lodewyk Wessels, Rene Bernards

Disclosure InformationRelationships relevant to this session

Tabernero, Josep

VHIO has received research funding from Agendia

Please note, all disclosures are reported as submitted to ASCO, and are always available at gicasym.org

Colorectal cancer different subtypes• Colorectal cancer is the second leading

cause of cancer death• Although several treatments exist we do

not have an optimal way to select treatments for individual patients

• Only KRAS status has been established as a predictor of anti-EGFR treatment activity

• New technology platforms allow genetic definition of different types of cancer based on gene expression and characterization

• Unbiased genome-wide analyses of gene expression patterns have been successful for molecular classification of BC & GBM

 

Molecular classification based on nearest centroid single sample predictor (SSP) 3 gene expression profiles (A, B & C)

Development Set (stage I-IV) (n=188) Netherlands Cancer Institute, Leiden Medical Center, Slotervaart

Whole Genome Array

Dev

elop

men

t

Validation Set (stage II-III) (n=543)Technical University Munich,

Institut Catala d’Oncologia & Vall d’Hebron Hospital Barcelona, Medical University Vienna, University of Ferrara

Valid

atio

n

Analysis of mutations in BRAF(V600), KRAS (codons 12, 13 & 61) and PIK3CA (exons 9 & 20) – All samples

Analysis of 615 (incl. kinome) by NGS – 73 samples

All

sets

Analysis of Epithelial and Mesenchymal genes – All samples

OS and Distant Metastasis-free survival (DMS) – All samplesEffect of adjuvant treatment – Stage III samples, validation cohort (n=123)

MSI analysis by IHC – All samplesMSI/dMMR gene expression pattern (64 gene signature1) – All samples

1Tian S et al. J Pathol. 2012

Development and Validation of the Molecular Subtype Signature

Development Cohort Validation Cohort

N 188 543Hospital NKI, LUMC, Slotervaart Vall d’Hebron, ICO

Barcelona, Munich Stage I

IIIIIIV

24 (13%)100 (53%)56 (30%)8 (4%)

-320 (59%)223 (41%)

-Gender F

M104 (55%)84 (45%)

226 (42%)317 (58%)

Subtype ABC

65 (35%)98 (52%)25 (13%)

117 (22%)336 (62%)90 (16%)

Unsupervised hierarchical clustering of whole genome reveals 3 distinct patient groups

Gene profiles were developed to identify these subgroups

A-Type B-Type C-Type

Unsupervised hierarchical clustering of whole genome reveals 3 distinct patient groups

Gene profiles were developed to identify these subgroups

A-Type(32 genes)

B-Type(53 genes)

C-Type(102 genes)

PROGNOSIS, MSI AND BENEFIT FOM ADJUVANT CHEMOTHERAPY

Clinical Characterization

Subtypes are significantly associated with prognosis

Risk of Distant Metastasis Risk of Death

C-TypeB-TypeA-TypeD

M R

isk

Dea

th

Subtypes are significantly correlated with benefit from adjuvant 5-FU-based treatment

Difference in proliferation between subtypes might explain difference in treatment benefit• significantly reduced expression of Ki67 and AURKA in C-type

compared to A- and B-type – Ki67 p=6.06e-5, AURKA p=4.53e-6

C-TypeB-TypeA-Type

Subtypes differ significantly in mutation and MSI frequency (Mismatch Repair deficiency)

• Cancer kinome sequencing (~600 kinases and other cancer related genes)– high mutation frequency in A and C-type (dMMR)– B type represent proficient mismatch repair (pMMR)

Subtypes differ significantly in mutation and MSI frequency (Mismatch Repair deficiency)

• Cancer kinome sequencing (~600 kinases and other cancer related genes)– high mutation frequency in A and C-type (dMMR)– B type represent proficient mismatch repair (pMMR)

Types Mutated genes MSI/dMMRA 36% 68%

B 17% 1%

C 34% 36%

Observed mutations in the cancer kinome

EPITHELIAL VS. MESENCHYMAL

Biological characteristics

Epithelial-Mesenchymal Transition

EMT markers are differently expressed in subtypes• Mesenchymal markers (higher in C-type)

– VIM, CDH2, FN1, FGFR1, FLT1, TWIST1, AXL, TGFB1• Epithelial markers

– CDH1, CDH3, CLDN9, EGFR, MET • Mesenchymal Character of C-type was confirmed by EMT signature

developed at MDACC (Loboda et al. 2011)

CONCLUSIONMolecular Subtypes

Biological featuresClinical featuresSubtype Clinical utility

Chemotherapy

New targeted therapy?(companion Dx)

No adjuvantor 5FU

Colon molecular subtype model

AcknowledgementsAll collaborators and patients

• Vall d'Hebron Hospital• Agendia• Institut Català d'Oncologia• Technische Universität Munich• Netherland Cancer Institute• Slotervaart Hospital• Leiden Medical Center• Medical University of Vienna• University of Ferrara• COLTHERES, EU-FP7