Current state of
breast cancer
classification
Marcella Mottolese
UOC Anatomia Patologica
screening programs and adjuvant
therapy changed the management of
breast cancer
impact in outcome with a
decrease in mortality in
most of the western world
despite increasing breast
cancer incidence
OPEN QUESTIONS
?
How to identify patients with node negative
disease at very low risk of relapse for whom the
risk/benefit ratio might be in favour of
chemotherapy?
How to predict the response to currently
available cytotoxic chemotherapy ?
How to identify tumor targets for directed
therapies?
Response to the adjuvant chemotherapy , mainly in
early stage of the disease, are affected by a complex
interplay of factors
Prognostic factors in breast cancer
Biomarkers Categories
1. Those that predict relapse or progression
independent of future treatment effects
designate as PROGNOSTIC FACTORS
2. Those that predict response or resistance
to a specific therapy designate as
PREDICTIVE FACTORS
Routinely used clinico-pathological
parameters
Age
tumor Size
lymph nodes
lympho-vascular
invasion
ER, PR; HER2;
Ki67
Tumor grade
Clinico-pathological
(tumor burden)
Biological
(intrinsic chacteristics)
Are these routinely used parameters
sufficient for individualized therapy
NO, because Brest Cancer is
HETEROGENEOUS
Clinically
Biologically
This clinical heterogeneity is driven by the genetic variability
of patients and tumors
A continuum of abnormal
gene expression predicts:
the tumorigenic
phenotype
the sensitivity of
tumors to treatment
Clinical
tools
St Gallen 2009 : news and progress
Genetic predisposition
Circulating tumor cells
Whole-genome studies
Stem cells
microRNAs
Networks in cellular
system
Multigene
assays
Angiogenesis
Pharmacogenetics
Resistance to
treatment by
crosstalk
INTEGRATING MOLECULAR AND
OTHER PATHOLOGICAL FEATURES
St. Gallen 2009
Endocrine
Responsive
(ER ≥ 1%)
Endocrine
Non-Responsive
(ER = <1%)
HER2-
Endocrine Therapy
(± CT using
suggested criteria)
CT
HER2+
Endocrine Therapy
+ Trastuzumab
+ CT
Trastuzumab
+ CT
St Gallen 2011
 Focus on the identification of “tumor
subtypes” to plan therapy more
Tumor
Subtypes defined
molecular classification
starting
Immunohistochemical
molecular
from
characterization
classification surrogate
Morphological evaluation of special
types
accurately
as
histological
Biology-driven trials
Andre F et al
2011 JCO
How can molecular tools help standard
pathology?
• Molecular sub-grouping
• Prognosis/Stratification
• Predictive markers
• Functional pathway read-outs
Standardization,
reproducibility & qualification
Molecular sub-grouping
….very distinct gene
expression patterns
irrespective of stage
(size, nodal, status) !
Sotiriou C, Pusztai L. N Engl J Med 2009;360:790-800
St Gallen 2011
‘Subtype’
Surrogate IHC markers
Type of therapy
Notes
Luminal A
ER and/or PgR + /HER2 -,
Ki-67 <14%
Endocrine therapy alone
Few require CT
(e.g. high nodal status).
Luminal B (HER2 -)
ER and/or PgR +/HER2-,
Ki-67>14%
CT
+ endocrine therapy
Inclusion and type of CT may
depend on perceived risk and
patient preference.
Luminal B (HER2 +)
ER and/or PgR +/HER2 +
CT
+ anti-HER2
+ endocrine therapy
HER2 +
(non luminal)
CT
+ anti-HER2
Triple negative (ductal)
CT
Consider DNA disrupting
agents.
‘Special histological types’
A. Endocrine responsive
B. Endocrine non responsive
Endocrine therapy
CT
Medullary and adenoid
cystic carcinomas may not
require any adjuvant CT.
No data are available to
support the omission of CT in
this group.
Ki67 (IHC)
British Journal of Cancer (2007)
Ki-67 prognostic cut-off point for all breast cancer subtypes
(multivariate analysis)
The best cut-off point with the lowest p-value and
highest HR was found at the Ki-67 index of 20%
Claudin-low breast cancer are characterized by:
 low to absent expression of luminal
differentiation markers,
 high enrichment for epithelial-to-mesenchymal
transition markers, immune response genes and
cancer stem cell like features.
Clinically, the majority of claudin-low tumors :
 are triple negative invasive ductal carcinomas
with a high frequency of metaplastic and
medullary differentiation.
 have a response rate to standard preoperative
chemotherapy intermediate between that of
basal-like and luminal tumors.
 resembles the mammary epithelial stem cell.
Prognosis/stratification
Distinct
Clinical outcome
Breast cancer should be considered according to
Hormonal Receptor, HER2, Ki-67
Disease free survival according to molecular subtype in 1015
breast cancer patients treated at IRE between 2000-2006
P-value (log-rank test) <0,0001
Estrogen receptors α and β show different associations to
clinicopathological parameters and their co-expression might predict a
better response to endocrine treatment in breast cancer
S Borgquist J Clin Pathol 2008;61:197-203
934 prospective
breast cancer
patients
To replace or to complement the
traditional biological parameters?
This may be highly misleading, because the
molecular classes are heterogeneous and
encompass different tumor types with
different risk profiles and different
responsiveness to the therapy
Heterogeneity may be evident in breast cancer
with basal-like profile since:
Not all the tumors with the basal-like profile are high risk
tumors. This molecular class also includes:
low-grade metaplastic carcinomas
adenoid-cystic carcinomas,
medullary carcinomas
low grade apocrine carcinomas
which have a very favorable prognosis
Predictive markers
ER and HER2
Breast Cancer
HER-2 neu
Negative predictive value
HIGH 95%
(<5% change to respond to antiestrogens or trastuzumab)
ER
Positive predictive value
30-50 %
a
b
30μ
d
c
30μ
10
μ
10
μ
a
b
b
What about gene
Expression signatures ?
The Role of Single Gene Analyses and Multigene Assays in Addressing
the Controversy on Chemotherapy or Not when Stage/Biology are
Discordant
21 Gene recurrence
Score (RS)
70 Gene
High Risk
Relative
Endocrine
“Resistance”
Relative
Chemo
“Sensitivity”
Three Strategies for the Development
of a Gene-Expression Prognostic Signature
1. “top-down” approach: geneexpression data from patients
with known clinical outcomes are
compared to identify genes that
are associated with prognosis
without any a priori biologic
assumption.
2. “bottom-up” approach: geneexpression patterns associated
with a specific biologic phenotype
or a
deregulated molecular
pathway are first identified and
then correlated with the clinical
outcome.
3. candidate-gene approach:
selected genes of interest on
the basis of existing biologic
knowledge are combined into a
multivariate predictive model.
Summary of multi-parametric tests for breast cancer
Ongoing RCT’s of prediction of
chemotherapy by multi-parameter assays
21 gene Recurrence Score Assay: Strongly
Predictive in NSABP B-20 (ER+No)
Mammaprint TM adds prognostic information in pT1 tumors
N=965 patients wit T1a,b,c tumors; adjuvant systemic treatment in 41%; median 7 years
Distant Metastasis-Free Survival
PT1ab
PT1c
T1A,BNoMo: ONCOTYPE DX
and MAMMAPRINT
TM
prolifiles
1 st generation signatures add prognostic information
to current “best”clinical tools
Breast Cancer Gene Expression Signatures and
Cell proliferation
70 gene signature
van’t Veer et al,2002
76 gene signature
Wang et al, 2005
OncotypeDX RS
Paik et al, 2004
Genomic grade index
Sotiriou et al, 2006
70
76
16
97
Cell cycle, angiogenesis,
invasion and metastasis
Cell cycle, proliferation, DNA
repair, immune response and
apoptosis
Proliferation, ER and HER2
invasion
Cell cycle and proliferation
44
Learning about adverse biological features
of T1a,bNoMo breast cancers
46
Stromal Clusters Reflect Different Microenvironments
Stromal Gene Expression Identifies Distinct Tumor Subclasses
Epithelia
Stroma
Stromal Subtypes are Associated with Outcome
47
The tumor microenviroment plays a key role in tumor
progression
Better understanding of intra-tumor diversity and epithelialstromal cell interactions will lead to more efficacious cancer
treatment
48
Functional Pathway
PI3K pathway gene signatures
Best way to develop these signatures
A PIK3CA-GS developed from PIK3CA mutant ER+breast
cancers
 PIK3CA mutations produce
characteristic change in
gene expression
 The mutant and the
associated gene signature
were associated with a
good prognosis in TAMtreated ER+/HER2- BC
Loi et al, PNAS 2010
121 breast cancer patients CMF treated
Unfavourable Biologic Profile (UBP≥3 ABF)
versus
Favourable Biologic Profile (FBP<3 ABF)
P-AKT
PI3K
Novelli F, Mottolese M SABCS 2009
Ki-67
HER2
TAKE MESSAGES
A gene signatures may be useful as a functional
read-out of pathway activity
Gene signatures will probably be most useful in
combination with genotype, ER, HER2, Ki-67
All predictive gene signatures need clinical
validation in appropriate clinical trials