molecular classification of breast cancer · 2015. 11. 27. · ck8/18 ck8/18 er, pr basal luminal...
TRANSCRIPT
Molecular classification of breast cancer – hormonal
receptors and beyond
Aleš Ryška
The Fingerland Department of Pathology
Hradec Králové
Can we do better???
Czech Republic
Hradec Králové
Concept of targeted therapy
Heinemann V, et al. Cancer Treat Rev 2013 DOI 10.1016/j.ctrv.2012.12.011.
“The ultimate goal of personalised medicine
is to define a disease sufficiently to enable
identification and treatment of only those
patients most likely to respond“
ONCOLOGIST PATHOLOGIST
Pathologist – integral part of multidisciplinary team
• morphology
• immunohistochemistry
• molecular biology
HER-2/neu & CEP 17
Diagnosis
TNM
Grade
Predictive markers
Breast cancer
• Heterogous group of tumors
– variable etiology, pathogenesis, prognosis, morphology, behaviour, response to treatment
progenitor cell
intermediary glandular cell
luminal cell
intermediary basal cell
myoepithelial cell
CK5/6, CK14 p63
CK5/6, CK14 P63, SMA
CK5/6, CK14 P63, SMA,
CD10
CK5/6, CK8/18
CK8/18 ER, PR
Luminal Basal
CK8/18 HER2
HER2+
ER
PR
HER2
Luminal A
DUCTAL LOBULAR
HER2+
Triple negative
Luminal B
ER+
PR+ Luminal (A or B)
HER2-
artefER-
artefPR- Triple negative !!!
HER2-
Sources of variability in tissue testing
Time to fixation Tissue processing
Type of fixation
Equipment
Quality of the
lab
Length of
fixation
Type of test
Experience
Demasking
Reagents
Controls
Computer
image
analysis
Interpretation
Scoring
Variation of
results
Post-analytic
phaze
Pre-analytic
phase
Analytic
phase
Wolff AC et al. J Clin Oncol 2007; 25: 118-145
Am J Clin Pathol 2010;134:813-819
Fixation
Operation
Modern Pathology (2009) 22, 1457–1467
30‘
120‘
Am J Clin Pathol 2010;134:594-596
… only cases known from core biopsy testing to exhibit 3+ overexpression of HER2/neu. Also, our estrogen receptor study3 included only high expressors …
Impact of fixation on Ki67 demonstration
Hitchman E et al. Histopathology. 2011 Dec;59(6):1261-3
Heterogeneity of neoplastic population!
Dowsett M, et al. J Natl Cancer Inst 2011;103:1656–1664
Varga Z, et al. (2012) PLoS ONE 7(5): e37379. doi:10.1371/journal.pone.0037379
time
Success of a biomarker
St. optimisticum
St. criticum
St. tragicum
Use common sense!
6980 patients 4651 ER+PR+ (66.6%) 1758 ER-PR- (25.2%) 481 ER+PR- (6.9%) 90 ER-PR+ (1.3%)
2.1%
D’Alfonso T: Am J Surg Pathol 2010
D’Alfonso T: Am J Surg Pathol 2010
HERretest project
Reference lab ISH
Reference lab IHC Negative Borderline Positive Total Discrepant
0,1+ (negative) 514 (83.3%) 13 (2.1%)2 8 (1.3%)2 535 (86.7%) IHC-/ISH+
N=21 (3.4%)
2+ (borderline) 57 (9.3%)3 5 (0.8%) 7 (1.1%) 69 (11.2%) IHC+/ISH-
N=59 (9.6%) 3+ (positive) 2 (0.3%)3 4 (0.6%) 7 (1.1%) 13 (2.1%)
Total 573 (92.9%) 22 (3.6%) 22 (3.6%) 617 (100%) Discrepant total
N=80 (13.0%)
617 cases – originally HER2 IHC negative (0, 1+) false negativity of primary testing = 4 % IHC-/ISH+ discordant phenotype = 3,4%
Clinico-patological features analyzed in the series
Feature Discordance
IHC- ISH + p1 OR (95% IS) p2
Total 3.4%
Age (years) - - 1.054 (1.013; 1.096) 0.009
Age >65 years 5.6 % 0.105 2.093 (0.873; 5.019) 0.098
Grade 3 4.3 % 0.617 1.355 (0.536; 3.425) 0.521
Tumor type DCI, PAP, other
(all but LCI, MUC, TUB) 4.9 % 0.011 3.695 (1.438; 9.499) 0.006
Tumor size > 20 mm 6.6 % 0.045 2.502 (1.040; 6.018) 0.041
ER: 0 % 5.7 % 0.116 1.960 (0.741; 5.188) 0.175
PR: 0 % 5.7 % 0.074 2.168 (0.894; 5.258) 0.087
Ki67 > 10% 6.1% 0.360 1.668 (0.602; 4.625) 0.326
Primary IHC 1+ 4.3% 0.356 1.607 (0.671; 3.853) 0.287
ER – and/or PR - 5.4 % 0.036 2.481 (1.032; 5.966) 0.038
Selection of IHC- cases advisable for ISH testing
Score N Discrepancy OR (95% IS) p1
0 60 0.0% Reference value (OR=1)
1 317 2.2%
2 189 4.2% 2.336 (0.834; 6.543) 0.106
3 51 11.8% 7.048 (2.269; 21.893) 0.001
• Histology – tumot type other than LCI, MUC, PAP • Tumor size > 20 mm (pT2 and higher) • Hormonal (partial or entire) independence – ER- and/or PR-
Koninki K, et al. HER-2 positive breast cancer: decreasing proportion but stable incidence in Finnish population from 1982 to 2005. Breast Cancer Research 2009
HER2 positivity in Czech population - 2014
90,0
87,2
86,6
88,0
86,6
86,6
86,9
4,9
3,8
3,9
10,0
8,9
8,5
8,2
10,0
8,7
9,2
4,6
3,3
4,0
0% 20% 40% 60% 80% 100%
% of patients
Negative
IHC 0/1+/2+ & ISH ≥ 2,0
IHC 3+
FN Olomouc a LF UP - Ústav
patologie
FNHK - Fingerlandův ústav
patologie
VFN Praha - Ústav patologie
MOÚ - Oddělení onkolog. a
experiment. patologie
Plzeň - Bioptická laboratoř s.r.o.
BIOLAB Praha, k. s.
Total
N = 20
N = 936
N = 508
N = 814
N = 2371
N = 1227
N = 5876
N = 5876
patients
Why are our predictions not 100% perfect?
Wu VS et al. J Steroid Biochem & Molec Biol 153 (2015) 45–53.
Cortazar R, et al. Lancet 2014; 384: 164–72
The Breast 23 (2014) 188e192
HER2
HR (ER/PR)
HER2 + HR
Normal conformation of HER2 molecule
Corrupt conformation of HER2 molecule
Arribas et al. Clin Cancer Res, 2010
Sperinde et al, Clin Cancer Res 2010
Predictive value of p95 protein
Can we do even better?
Advances in genetics a proteomics
• High resolution SNP chips • array CGH • multiplex PCR • deep parallel sequencing • gene expression profiling • .....
• Shift from research to practice • Costs of these methods are dropping down • Great expectations
Class discovery /molecular taxonomy
• Identification of specific molecular subclasses, having biologic and clinical significance
– Perou et al., Nature, 2000
– Sørlie et al., Proc Natl Acad Sci USA, 2001
• 42 patients
– (36 DIC, 2 LIC, 1 DCIS, 1 FA, 3 x normal tissue)
• ,,Hierarchical clustering“ of samples based on similarity of gene expression
• 1753 genes studied
• expression in individual sample compared to median of expression of the same gene in all samples, color coded
• Subsequent analysis based on expression of ,,Intrinsic gene subset“ – 496 genes
• gens selected to ensure that paired samples of the same tumor would be always classified together and would differentiate as much as possible from the remaining tumors
The number of clearly different molecular phenotypes observed among the breast tumours suggests that we are far from having a complete picture of the diversity of breast tumours. When hundreds (instead of tens) of breast tumours have been characterized, a more precisely defined tumour classification is likely, and statistically significant relationships with clinical parameters should be uncovered….
• 85 samples
• 78 carcinomas (including 40 from previous study)
• Overall and relapse-free survival analysis of the 49 breast cancer patients, uniformly treated in a prospective study
A
B
Class discovery /molecular taxonomy
Alternative (simpler and cheaper) approaches • molecular taxonomy may be simulated by e.g.
clustering analysis of immunohistochemically detected proteins
• immunohistochemical detection of 26 proteins
• 552 patients with early BC
• more than 1600 samples (tissue microarrays)
• analysis of interobserver agreement in data interpretation among five experts
• 3 cohorts of BC (799 cases)
• clustering analysis using 5 different “intrinsing gene“ sets
• consensus among 5 experts: κ= 0,435 – 0,757
• perfect agreement (5/5) in 42,4 -63,6 % cases
• leaving luminal type out (A,B,C) κ= 0.435–0.582
• perfect agreement (5/5) in 17.5- 46.1% cases
• κ=-1 perfect disagreement
• κ= 0 consensus fully random
• κ= 1 perfect agreement
• Κ=0.01–0.2 minor agreement
• κ= 0.21–0.4 slight agreement
• κ= 0.41–0.6 average agreement
• κ= 0.61–0.8 substantial agreement
• κ= 0.81–1.0 (nearly) perfect agreement
Clin Cancer Res 2006;12:781-790.
• 633 cases - expression of 5 “traditional markers“ (ER, PR, Ki67, HER2, p53) 4 clusters
Class prediction
• Several multigene predictors (gene signatures) exist, predicting behaviour and tratment response
– minimum of common genes
– similar performance – similar, but not identical prognostic groups
– majority built and verified on specific cohorts of BC (e.g. ER+/N0)
• OncotypeDX (21 genes) – ASCO / NCCN guidelines
• RT-PCR (FFPE material)
• Recurrence Score: low – intermediate - high
• ER+ a N0 treated by tamoxifen; other groups... • basically – objectification of proliferation in ER+ BC
• MammaPrint (70 genes) – FDA
• oligonucleotide microarray (fresh tissue), transport in medium within 5 days to central laboratory
• good prognosis / poor prognosis group
• N0 < 60 years < 5 cm TU
• BC Gene Expression ratio, Veridex LLC, Endopredict
• predictors of response to CHT, hormonal, targeted Th
Class discovery /molecular taxonomy
• Enormous expectations: taxonomy should bring dramatic improvement of clinical management
• Reality:
– Validation ???, reproducibility ???, sets of genes, number of classes clinical implications ???
• e.g. majority of luminal tumors are HR+ (routinely identified by IHC) • any other subclassification based on proliferation is arbitrary
• What about luminal cases which are HR- ?
• HER2+ BC is managed the same way irrespective of molecular subclass
• molecular taxonomy is in fact reflecting expresion of HR, HER2 and proliferation
Class prediction
• similar problems with standardization, quality control and clinical validation
• What is the ,,added value“ compared to classical molecular markers (ER,PR, HER2, Ki67) ?
• 40-60% clinically intermediate cases (grey zone) remains unresolved also after OncotypeDX
• first randomized clinical trials (OncotypeDX-
TAILORx, MammaPrint - MINDACT)
Conclusions
• Individual molecular markers (ER, PR, HER2...) are used for decades and represent currently a standard of classification and patients stratification for individual treatment
• advances in molecular genetics a proteomics enabled new possibilities of molecular classification and prediction in BC
• however, despite great expectations these methods do not replace traditional classification, they are oversimplifying complexity of the BC spectrum
• nevertheless, they can be a significant addition in prognosis determination and prediction of treatment effect, namely in selected groups and in „grey zone“ cases not resolved by classical approach
???
Hanahan D, Weinberg RA. Cell. 2011; 144(5): 646-74.
Brychtova S. et al.: Stromal Microenvironment Alterations in Malignant Melanoma, DOI: 10.5772/22912
Lawrence MS et al. Nature 2013; 499: (7457):214-8.
Impact of gene mutation on expression of encoded protein
Hornychova H, et al. Tumor-infiltrating lymphocytes predict response to neoadjuvant chemotherapy in patients with breast carcinoma. Cancer Invest. 2008; 26: 1024-31.
Denkert C, von Minckwitz G, et al. J Clin Oncol. 2015 ;33(9):983-91.