Deletion of chromosome 11q predicts response to anthracycline-based chemotherapy in early breast cancer

Despite recent consensus on eligibility of adjuvant systemic therapy in lymph-node negative breast cancer (NNBC) patients based on clinico-pathological criteria, specific biological markers are needed to predict sensitivity to the different therapeutic options. We examined the feasibility of developing a genomic predictor of chemotherapy response and recurrence risk in 185 patients with NNBC using assembled arrays containing 2,460 BAC clones for scanning the genome for DNA copy number changes. After surgery, 90 patients received anthracycline-based chemotherapy whereas ninety-five did not. Tamoxifen was administered to patients with hormone-receptor positive tumors. Association of genomic and clinico-pathological data and outcome was computed using Cox proportional hazard models and multiple testing adjustment procedures. Analysis of NNBC genomes revealed a common genomic signature. Specific DNA copy number aberrations were associated with hormonal receptor status, but not with other clinico-pathological parameters. In patients treated with chemotherapy, none of the genomic changes was significantly correlated with recurrence. In patients not receiving chemotherapy, deletion of eight BAC clones clustered to chromosome 11q was independently associated with relapse (DFS at 10 years±SE, 40±14% vs. 86±6%;p<00001). The 54 patients with deletion of 11q (29%) did not present more aggressive clinical-pathological features than those without 11q loss. The adverse influence of 11q deletion in clinical outcome was confirmed in an independent validation series of 88 NNBC patients. Our data suggest that NNBC patients with 11q deletion may benefit from anthracycline-based chemotherapy despite other clinical, pathological or genetic features. However, these initial findings should be evaluated in randomized clinical trials. Keywords: CGH, breast cancer STUDY DESIGN AND SELECTION OF PATIENTS From September 1979 to June 2000, over 3.100 new breast cancer patients were diagnosed at the University of Valencia. Of them, 1.482 tumors were criopreserved at -80ºC. For this study, biopsies were randomly selected based on the following criteria for inclusion: (1) diagnosis of primary invasive breast carcinoma of any size; 2) treatment by modified radical mastectomy or breast-conserving surgery, including dissection of axillary lymph nodes, followed by radiotherapy if indicated; 3) the apical axillary lymph nodes were tumor-negative (pathological examination, pN0); and 4) complete clinical data were available. Over 363 samples fulfilled the criteria. Previous to DNA extraction, these frozen tumor sections were stained by hematoxylin/eosin (H&E) and reviewed for tumor infiltratrion: only those with clear >50% of tumoral cells were selected. A cohort of 185 patients fulfilled these final criteria. Clinico-pathological variables including tumor size, histological grade and subtype and ER and PR status were determined following standard methods as reported.(26) Human investigations were performed after approval by an institutional review board on scientific and ethical affairs. MICROARRAY-BASED COMPARATIVE GENOMIC HYBRIDIZATION (ARRAY CGH) DNA extraction, hybridization and imaging. Frozen tumors were included in OCT compound. Previous to DNA extraction, H&E stained tumor sections were examined to select samples with more than 50% of tumoral cells. Around 20-30 sections of 25Nm were used for DNA extraction. After removing the OCT with PBS washes, DNA was extracted as previously described.(26) Genome-wide analysis of DNA-copy number changes was performed using array CGH on a microchip with 2.460 BAC and P1 clones printed in triplicate (UCSF Hum Array 2.0) with a resolution of 1.4 Mb across the genome.(27) Methods and analytical procedures have been described elsewhere in detail (25, 27, 28) 6 (Figure 1). For visualization of genomic data, the TreeView program 1.60 (Stanford, CA) was used. To confirm array CGH data, CGH to chromosomes was performed in 44 biopsies included in the study. Interphase FISH analysis. To confirm specific gains and losses of BAC clones observed in the array CGH analyses, fluorescence in situ hybridization (FISH) studies using individual BAC clones as probes on isolated nuclei from frozen tumor sections was performed using a reported technique.(29) The gene loci examined corresponded to 5 overrepresented and 4 deleted BAC clones, using appropriate centromeric probes as controls. A total of 100 cells were examined on each of the 22 tumors examined. These clones were obtained from RZPD German Resource Center (Berlin, Germany) or purchased from Vysis (Downers Grove, IL, US). STATISTICAL DATA ANALYSIS To process the genomic data obtained with array CGH we used a previously described analytical model.(19, 30) Arrays were normalized by subtracting the median of each array from the average log2 ratio for every clone. The array CGH data were then analyzed using Hidden Markov Model (HMM) using the Bioconductor package aCGH. Log2 ratios as ordered in the genome were segmented into regions of constant copy number. In addition, the HMM model was employed to impute missing values by using the estimated copy number ratio for the segment containing the clone(s) with missing values. Clones with missing values located between segmented regions were assigned the mean value of the segment that is closer in genomic distance. Thus, each clone was assigned a segment value referred to as its "smoothed" value. Median absolute deviation (MAD) of the difference between the observed and smoothed values was used to estimate the tumor-specific experimental variation. Clones for each array were assigned into three groups: gained (smoothed log2 ratio >3 times the MAD); lost (smoothed log2 ratio <3 times the –MAD); and not changed (the log2 ratio cannot be assigned to lost or gained groups). Smoothed data was used to study association with the following phenotypes: age, tumor size, histological grade and subtype, stage, estrogen and progesterone receptor status, and recurrence/survival. Finally, we used Fisher’s two-sided exact test 2x2 crosstabs to compare genomic events or clinical variables among both groups of treatment. To evaluate differences in disease-free survival, Kaplan-Maier survival curves for the sets of patients were examined. Gene Ontology (GO) classification. GO is used to construct a unified and structured vocabulary for the description of genes and their products in any organism. A GO term is descriptive of the molecular function, biological process or cellular component. We then performed a statistical analysis to check if there were any GO categories that were enriched in the genes located in the region of deletion in chromosome 11q23-q24 with respect to the whole genome. CLINICAL SERIES FOR VALIDATION OF ARRAY CGH RESULTS To validate the possible association of chromosome 11q deletion with increased relapse rate, this was tested in a validation group of 88 tumor biopsy samples from an independent cohort of NNBC patients. These were 18 Spanish patients treated in different Institutions within the Valencia area whose genomes were analyzed with array CGH as described above. In addition, data from 70 patients were obtained from a recently published series of American breast cancer patients analyzed using similar whole-genome array CGH techniques (Chin et al, in press). All patients fulfilled the reported inclusion criteria of the study. Kaplan-Maier survival curves for the two sets of patients were evaluated. Clinicopathological characteristics of the validation series are shown in Supplemental Table S1.