Technology Transfer in Diagnostic Pathology. 6th Central European Regional Meeting. Cytopathology. Balatonfüred, Hungary, April 7-9, 2011. RECENT ADVANCES IN THE MOLECULAR DIAGNOSIS OF BREAST CANCER Philippe Vielh MD, PhD Breast cancer study group Institut de cancérologie Gustave Roussy Villejuif, France
OUTLINES Local and overall context for FNAC FNA material for molecular studies First approach Second approach Perspectives
LOCAL AND OVERALL CONTEXT FOR FNAC Fine needle aspiration cytology (FNAC) is extensively used in the diagnosis of breast lesions as it is a simple, minimally invasive and cost-effective method However, depending on the local context, it is not used by the clinician because the rates of unsatisfactory specimens and/or false negative and/or suspicious results are too high
Benign or malignant?
Benign or malignant?
Cytological results at our one-stop clinic for breast nodules In a series of 1824 lesions (1741 pts) sampled by FNAC (April 2004-March 2007) Unsatisfactory specimens : 3% Suspicious : 8% Cancer : 46% Benign : 43% Cancer : 80% Fx positive: 0.1% Fx negative : 3%
Cytological diagnosis Question: can we do it better? How? Experience Immunocytochemical methods Molecular studies (expression arrays) Is FNA an adequate technique for obtaining material for subsequent identification of pertinent biomarkers (diagnosis, prognosis and prediction)?
FNAC MATERIAL FOR MOLECULAR STUDIES Quantity Quality Concordance
Patients and methods Prospective series of breast lesions sampled by FNAC Immediately processed for diagnostic and research purposes with written informed consent of the patients Gold standard of the cytological diagnostic : histopathology and outcome for several benign cases
Characteristics of the population Benign (n=39) Malignant (n=75) Median age (years) 47 (18-85) 60 (30-92) Palpable lesion 48.7% 78.3% Median size (mm) 15.8 (5-35) 26.7 (6-110) Birad ACR 4 25.6% 20.3% Birad ACR 5 2.6% 79.7% FNAC classified as malignant by the cytopathologist FNAC classified as suspicious by the cytopathologist 2.6% 92% 7.7% 6.7%
Materials and methods RNA extraction : - Qiagen kit - cdna obtained from 600 ng of RNA - RNA quantification Quantitative PCR analysis (Taqman)
Quantity and quality of RNA 144 FNA 114 mrna with measurable quantity 111 samples with >1µg mrna (97%) 109 samples evaluated for quality (RIN) 97 OK for arrays (89%) Median RIN of all samples : 7.5 (1.7-9.5)
Correlations between ER levels assessed by PCR and IHC 30 25 ESR1 (dct) 20 15 10 5 0 ER (IHC) Correlation : 94% Uzan C et al, Cancer 2009;117:32-9
Correlations between HER2 assessed by PCR and IHC (+/- FISH) 20 ERBB2 dct 18 16 14 12 10 8 6 4 2 0 ERBB2 IHC Correlation : 98% Uzan C et al, Cancer 2009;117:32-9
First approach for selecting biomarkers of diagnosis Using a pangenomic array (44K, Agilent) Selection of candidate genes up or down regulated (x2) from a previous study comparing 148 breast cancers with a pool (Clonotech) of breast normal tissue (collaboration with the Transbig consortium) 666 genes identified (435 with known functions)
Series 1: exploratory FNA material: 13 adenocarcinomas et 10 benign lesions mrna expression of the 22 top genes found in the Transbig profile 3 housekeeping genes : gene expression evaluated by real time QPCR Normalisation : mean of 3 housekeeping genes (18S, GAPDH, RLPO) Univariate analysis using the Student t test
Series 1: exploratory Univariate analysis, 6 genes with a P value<0.001
Series 1: conclusions Four out of the six genes (CENPF, MMP11, KTNC2, FN1) were highly predictive of diagnosis (all four with p<10e-7). In the multivariate analysis, the best-fit model comprises two genes: CENPF (p=0.001) and MMP11 (p=0.003) The c-value of the two-gene model is 0.96 [0.92-1.00]
Series 2: validation Odds Ratio Estimates Point 95% Wald Effect Estimate Confidence Limits CENPF 0.583 0.309 1.099 p=0.10 MMP11 0.386 0.221 0.671 p=0.0008 C VALUE OF CENPF +MMP11 model is 0.92 (95% CI [0.86-0.98])
Conclusions of the 1st approach High reproducibility of the results But few genes included in the diagnostic profile And not better than cytomorphology!! How can do it better?
Second approach for selecting biomarkers of diagnosis Gene expression arrays have generated molecular predictors of relapse and drug sensitivity Aim: identify exons differently expressed in malignant and benign breast lesions and build a molecular classifier for breast cancer cytodiagnosis by means of splice array
Alternative RNA splicing: one gene/several transcripts Genomic DNA Transcription Exon 1 Exon 2 Exon 3 Exon 4 Exon 5 Intron 1 Intron 2 Intron 3 Intron 4 Genome Unprocessed Transcriptome RNA Splicing A B Pre-messenger RNA Exon 1 Exon 2 Exon 3 Exon 4 Exon 5 Exon 1 Exon 2 Exon 4 Exon 5 Transcriptome Translation messenger RNA Proteome Function A New protein structure and function - Affects signaling pathways - Modifies drug efficacy Function B
Flow chart of the study
Results of the 2 nd approach 165 breast FNA: 120 adenocarcinomas and 45 benign lesions A molecular classifier with 1228 probe sets was generated from the training set of 94 FNA This signature accurately classified all samples (accuracy=100%, 95% CI: 96-100%) The molecular predictor accurately classified 68 of 71 lesions in the validation set of 71 FNA (accuracy=96%, 95% CI: 88-99%)
Correlation of each specimen with the average benign and malignant profiles
Expression of the 10 most overexpressed genes in breast cancer
Main pathways differentially expressed
Conclusions of the 2 nd approach Many exons are differentially expressed by breast cancer and benign lesions These alternative transcripts are detectable on material obtained by FNAC and may contribute to increase sensitivity and specificity of breast cancer cytodiagnosis
Anticipation for a diagnostic test
Benign or malignant?
Benign or malignant?
Thank you!! KÖSZÖNÖM!!