Frequency, prognostic impact, and subtype association of 8p12, 8q24, 11q13, 12p13, 17q12, and 20q13 amplifications in breast cancers

We studied a consecutive series of 547 unilateral localized invasive breast carcinomas from women treated at the Institut Paoli-Calmettes between October 1987 and December 1999. According to the WHO classification, this series comprised 386 ductal, 72 lobular, 37 tubular, 8 medullary carcinomas and 44 other histological types. They were obtained after informed consent and stored in an anonymous fashion according to an approval of the local Ethics Committee. The Ethics Committee of the Institut Paoli-Calmettes (Marseille's Cancer Institute) approved the use of these specimens and the data in research. The average age at diagnosis was 59 years (range 25–94). Raw survival data were either obtained from the cancer registry of the Institut Paoli-Calmettes or collected from the patients attending physicians. The pathologic stage, tumor diameter, and nodal status were obtained from the primary pathology reports. A total of 254 tumors were associated with lymph node invasion and 403 were positive for estrogen receptor. All slides from all tumors were reviewed by one of two pathologists (J. J. and E.C.J.) to define the various histoclinical factors collected for this series. They included patient age, invasive histological type, pathological tumor size, Scarff-Bloom-Richardson (SBR) grade (I to III), peritumoral vascular invasion, axillary lymph node status, estrogen receptor expression (ER), progesterone receptor expression (PR), P53 status, as evaluated by immunohistochemistry (IHC) with a positivity cut-off value of 1%, ERBB2 status, evaluated by IHC with the 0-3+ score as illustrated by the HercepTest kit scoring guidelines (DakoCytomation, Coppenhagen, Denmark), and Ki67 status as evaluated by IHC with a positive cut-off value at 20%. This study was approved and executed in compliance with our institutional review board.

Tissue microarray (TMA) was prepared as described previously [35]. Five-μm sections of the resulting TMA block were made and used for fluorescence in situ hybridization (FISH) and IHC analysis after transfer onto glass slides.

To characterize the 8q24 and 17q12 amplified regions, FISH on TMA was done according to the histology FISH instructions of DakoCytomation with the MYC probe and HER2 FISH pharmDx™ kit (DakoCytomation, Coppenhagen, Denmark). To characterize the 8p12, 11q13, 12p13, and 20q13 amplified regions, FISH on TMA was done according to published protocols [36, 37]. Two 8p12 subregions centered on RAB11FIP1 and FGFR1 loci, respectively, were analyzed. From telomere to centromere, the different BAC pools were constituted as follows: BAC pool 1 (RAB11FIP1 region): RP11-863K10 (AC138356; chr8:37,630,477-37,820,753), RP11-113G10 (chr8:37,715,876-37,881,776), RP11-457O21 (chr8:37,881,796-38,044,340); BAC pool 2 (FGFR1 region): RP11-90P5 (AC084024.17; chr8:38,091,019-38,226,422), RP11-513D5 (AC087362.13; chr8:38,216,677-38,358,846), RP11-100B16 (chr8:38,358,839-38,522,417), RP11-675F6 (AC069120.9; chr8:38,485,749-38,646,922). The 11q13 region centred on CCND1 was analyzed with the following combination of BAC pools, from centromere to telomere: RP11-300I6 (AP001888; chr11:69,162,462-69,323,966), RP11-643C9 (chr11:69,297,662-69,494,887), RP11-626H12 (AP003555; chr11:69,478,620-69,600,219). The 12p13 region spanning NOL1 region was analyzed with, from telomere to centromere, RP5-940J5 (AC006064; chr12:6,422,311-6,594,917), RP11-433J6 (AC135892; chr12:6,579,330-6,755,900), RP11-578M14 (chr12:6,699,832-6,884,088) BAC pool. Three 20q13 subregions of amplification corresponding to MYBL2, ZNF217 and AURKA loci were analyzed with locus-specific BAC pools, from centromere to telomere : MYBL2 locus BAC pool : RP11-69I10 (chr20:41,467,083-41,631,733), RP11-153L9 (chr20:41,659,456-41,808,516), RP5-1030M6 (AL035089; chr20:41,816,168-41,989,971); ZNF217 locus BAC pool : RP11-91L1 (chr20:51,421,217-51,572,829), RP4-724E16 (AL157838; chr20:51,561,511-51,690,363), RP11-299C12 (chr20:51,647,272-51,837,964); and AURKA locus BAC pool : RP11-380D15 (AL139824; chr20:54,122,911-54,316,054), RP5-1167H4 (AL121914; chr20:54,336,458-54,472,150), RP5-1153D9 (AL109806; chr20:54,472,051-54,566,171). Genomic information was taken from the UCSC Genome Browser on Human (March 2006 Assembly), which is based on NCBI Build 35 (National Center for Biotechnology Information, U.S. National Library of Medicine 8600 Rockville Pike, Bethesda, MD, USA).

DNA from BAC clones were purified, labeled and individually verified for their specificity of their addressed regions. All BAC clones were obtained from the BACPAC resource (Children's Hospital Oakland – BACPAC Resources, Oakland, CA, USA). After counterstaining with Vectashield containing 4,6-diamidino-2-phenylindole (DAPI) (Vector, Burlingame, CA, USA), images were analyzed with a microscope (DMRXA, Leica Microsystèmes, Marseille, France), captured with a CCD camera, filtered and processed with ISIS software (In Situ Imaging Systems, Metasystems Hard- und Software GmbH, Altlussheim, Germany)[38].

Fluorescence was scored on a minimum of 50 nuclei per tumor. The 50 nuclei of cancer cells were representative of the overall cell heterogeneity of the tumor. For each region analyzed, two observers read the TMA independently. The region analyzed was considered as amplified when the number of BAC pool signal was >5 in the cell. Tumors were defined as amplified when 10% or more of tumor cells showed such amplification.

The characteristics of the antibodies used are listed in Table 1. IHC was done as previously described [35], using LSAB2 kit in the autostainer (Dako Autostainer, Copenhagen, Denmark). Results were evaluated under a light microscope by two pathologists (EC-J, JJ) and scored by the quick score (QS) as previously done [35], except for ERBB2 status, which was evaluated with the Dako scale (HercepTest kit scoring guidelines). For each tumor, the mean of the score of a minimum of two core biopsies was calculated.

Amplification data were summarized by frequencies and percentages. Clinical data were dichotomized as follows: Amplicon: amplified vs. non-amplified, grade: I vs. II/III, age: <50 vs. ≥ 50, tumor size: pT1 vs. pT2/pT3, peritumoral vascular invasion: absent vs. present, estrogen receptor and progesterone receptor: negative vs. positive, Ki67: <20 vs. ≥ 20 and axillary lymph node: negative vs. positive.

The association between two categorical variables was examined using Fisher's exact or χ² tests. The primary endpoint was the metastasis-free survival (MFS), which was defined by the time interval between the diagnosis of breast cancer and a distant metastasis. Metastasis-free patients were right censored at the date of the last follow-up, death, recurrence of local or regional disease, or development of a second primary cancer. Survival curves were derived from Kaplan-Meier estimates and compared by log-rank test. Significant changes of relative risks of metastasis according to the amplification status were explored using Cox's proportional hazard models in univariate and multivariate analysis. Multivariate models were built using a backward stepwise selection of variables to minimize the Akaike Information Criterion. All results are presented with their 95% confidence intervals. Statistical tests were two-sided at the 5% level of significance. All the statistical analyses were done using R.2.3.0 statistical software [39].

Six regions of amplification were analyzed: 8p12 (amplified at least at one locus: RAB11FIP1 and/or FGFR1), 8q24 (MYC), 11q13 (CCND1), 12p13 (NOL1), 17q12 (ERBB2), and 20q13^Co (co-amplification of either two of three loci: MYBL2, ZNF217 and AURKA, or all three) regions. For the 20q13 region, in regard to the significant impact of ZNF217 amplification on disease evolution [29], we decided to distinguish the amplified subregion 20q13^Z centered on ZNF217. To assess the frequencies of the 8p12, 8q24, 11q13, 12p13, 17q12 and 20q13 amplifications we used FISH on TMA with specific probes. The number of interpretable cases varied between different FISH experiments (Figure 1). All TMA sections were only hybridized once. Reasons for non-informative results were: lack of tissue on the TMA, absence of unequivocal tumor cells, or non-interpretable hybridization data.

Among the informative cases, the frequency of amplification was 22.8% for 8p12, 6.1% for 8q24, 19.6% for 11q13, 4.1% for 12p13, 9.9% for 17q12, 9.9% for 20q13^Z and 8.5% for 20q13^Co regions (Figure 1). More in detail, the amplification frequencies of subregions centered on RAB11FIP1 and FGFR1 (8p12) as well as on MYBL2, ZNF217 and AURKA (20q13) loci are listed in Table 2. The highest amplication frequencies were obtained with RAB11FIP1 and ZNF217 subregions for 8p12 and 20q13, respectively.

We then looked at the distribution of the frequency of amplifications and co-amplifications. Overall, only 128 cases were informative for all the regions analyzed. A total of 57.4% of cases showed no amplification. The frequency of single amplifications was 9.8% for 8p12, 3.3% for 8q24 and 12p13, 1.6% for 20q13^Co and 20q13^Z. The 11q13 and 17q12 regions were never found amplified alone. The frequency of co-amplifications was 8.2% for 8p12/11q13, 3.3% for 11q13/17q12, 1.6% for 17q12/20q13^Co, 11q13/20q13^Co, 11q13/20q13^Z/20q13^Co, 8q24/11q13/17q12, 8q24/20q13^Z/20q13^Co, 8p12/11q13/12p13 and 8p12/11q13/20q13^Z/20q13^Co (Figure 2). We also looked for associations between amplified regions. The 11q13/20q13^Co, 12p13/20q13^Z and 8p12/11q13 co-amplifications were among the most strongly correlated [see Additional file 1].

We next examined the relation between amplifications and histoclinical factors [see Additional file 2]. We did not find any association between amplified regions and age, histological type and pathological tumor size. Amplification of 8q24, 12p13, 17q12 and 20q13^Co regions were correlated with high grade. Amplification of 8p12, 12p13 and 17q12 were correlated with a high proliferation index. Amplification of 8q24, 12p13, 17q12, and 20q13^Z were associated with ER and/or PR negativity. The 20q13^Co amplification was associated with axillary lymph node invasion.

We then examined the impact of amplifications and co-amplifications on clinical outcome. We did univariate analyses to determine the impact of each individual amplified region on the MFS. The 8p12 and 17q12 amplifications but not the others were associated with clinical outcome (Table 3). Both were associated with decreased five-year MFS in the whole population and in N- patients (Figure 3A–D).To estimate the impact of co-amplification on MFS we did a multivariate analysis of significant amplifications in univariate analysis followed by a stepwise selection. We found that co-amplification of 8p12/17q12 was associated with a poor outcome in N- patients (Figure 3E).

We further evaluated the importance of the amplifications as prognosis markers. We did a Cox multivariate analysis of MFS. The values for amplification, grade, age, tumor size, peritumoral vascular invasion, ER, PR, and Ki67 were considered as categorical variables. Amplification status of 8p12 remained significant as well as Ki67 status according to the Akaike Information criterium when dichotomized amplified vs. non-amplified and <20 vs. ≥ 20 in N- patients (Table 4). The relative risk of recurrence was 2.52 for 8p12-amplified disease compared to non-8p12-amplified disease (p = 0.15).

Five main molecular subtypes (luminal A, luminal B, basal, ERBB2-overexpressing, and normal-like) have been identified by gene expression profiling of breast tumor samples using an intrinsic set of ~500 genes [40, 41]. We determined the subtype of our samples by both mRNA and protein analyses (not shown). The 11q13 amplification was strongly correlated with the luminal A subtype (p value = 0.00029) and negatively correlated with the basal subtype (p value = 0.005633), whereas the 12p13 amplification was correlated with the basal subtype (p value = 0.020). As expected, the 17q12 amplification was correlated with the ERBB2 subtype (p value = 2.061e-06) (Table 5).

The most frequent amplification and co-amplification involved the 8p12 region. Several studies have shown that 8p12 is a common region of amplification and may harbor important breast cancer oncogenes [5, 44‐46]. This region, like other "hot spots" for gene amplification, may contain several genes that contribute to cell transformation. Because the exact 8p loci of significance are not known, we analyzed two subregions of 8p12 spanning many potentially relevant genes [5, 44‐46]. The two subregions covered genes that are amplified and most of them are overexpressed [5]. The frequency of 8p12 amplification in our breast tumor series was greater than the 10–15% commonly reported in literature, but similar to that recently reported by Garcia et al. (2005) [45]. Our results strengthen the idea that the 8p amplification, when all subregions are combined, may occur in a higher number of breast cancer than published so far. The 8p amplification was correlated with the amplification of 11q13 region: while 24.6% of tumors contained either amplified 8p12 region or 11q13 region, simultaneous amplification of both was seen in 10.8%. This result is in agreement with a previous study [47, 48]. This suggests a coordinated mechanism of amplification and oncogene activation, and some kind of relation between proteins encoded by genes from the two regions. The 8p12 amplification tended to have aggressive tumor features such as high proliferative index and high SBR grade, but no association with histological subtype was found. With respect to prognosis, the 8p amplification was a significant predictor of reduced MFS. Multivariate analyses indicated that amplification of 8p12 added to the prognostic power of Ki67 status to define high risk N- patients. Our findings are consistent with the idea that 8p12 is a common region of amplification that may play a role in tumor behavior and/or pathogenesis. Previous analysis of 8p12 subregional amplifications in breast cancer [5] pointed to the amplified FGFR1 subregion as the best prognostic marker of bad disease evolution among other 8p amplifications. The identification of the best candidate drivers or the best therapeutic targets could refine the impact of the 8p12 amplification on survival and help determine which population should be targeted.

The MYC gene on 8q24 encodes a transcriptional regulator whose expression is strongly associated with cell proliferation. MYC amplification occurs in several types of cancers [49, 50]. We found that the incidence of MYC amplification was in the range of those described previously [19]. About 13% of tumors contained either amplified MYC or ERBB2, whereas simultaneous amplification of both was seen in 1.9%. About 22% of tumors contained either amplified MYC or CCND1, whereas simultaneous amplification of both was found in 4.2%. MYC amplification was correlated with high grade and tended to be associated with high proliferation index, in agreement with previous studies [19]. Although the expression of the MYC protein is stimulated by estrogen and downregulated by tamoxifen in hormone-responsive breast tumors in vitro [51‐54], we found that MYC amplification was correlated with the absence of estrogen receptor, which is in agreement with other studies [21, 55] but contradictory to others [56, 57]. No correlation was found with hormonotherapy (data not shown). With respect to prognosis, MYC amplification was not associated with MFS, neither in the whole population nor in the two different lymph node populations of patients. This is consistent with a previous study [58].

Amplification of the 11q13 region is a relatively frequent event in breast tumors [59]. This region harbors four distinct subregions of amplification, which can be amplified independently or together in different combinations [59, 60]. Candidate genes have been suggested such as CCND1, EMS1, and PAK1 [60]. Amplification of CCND1 is within the most frequently amplified subregion and is found in two-thirds of all 11q13 amplifications. CCND1 overexpression promotes tumorigenesis in transgenic mice [61]. Amplification could promote sustained expression, which may cause the cell to cycle continuously. Amplification of CCND1 was found in 19.6% which is close to what is usually found in breast tumors. No association with histoclinical factors or survival was found, in disagreement with published data [22, 62].

Amplification of 12p13 in breast cancers was first identified by Dib et al. (1994) [6] and characterized by comparative genomic hybridization by Yao et al. (2006) [7]. Overexpression of presumptive amplicon genes was found in medullary breast cancers [25]. Amplification of the short arm of chromosome 12, mostly as isochromosome, or as amplified 12p11-12 and 12p13 regions, is a hallmark of testicular germ cell tumors (TGCT) [63‐65] and its overrepresentation is related to invasive growth of TGCT [66]. The 12p13 region harbors several candidate cancer genes. Human embryonic stem cell genes NANOG, GDF3 and STELLA, which are downregulated when cells commit differentiation, are expressed in TGCT and in breast cancers [67, 68]. This suggests a role of these stem cell genes in carcinoma progression. Like CCND1, the CCND2 protein is involved in G1 phase of the cell cycle. NOL1 encodes the nucleolar protein P120, a proliferation-associated antigen that is temporally regulated during the cell cycle with an increase in protein expression at the G1/S transition. The choice of the NOL1 region for FISH was further suggested because its protein expression is associated with overall survival in node-negative breast cancers [69]. However, we did not find any association between amplification and survival. The 12p13 amplicon was correlated with 20q13^Z, which may also be associated with proliferation.

ERBB2 amplification was found in 9.9% of tumors and was associated with high grade, absence of steroid receptors, overexpression of ERBB2 protein, and high proliferation index, which is an agreement with previous studies [17, 70, 71]. In univariate analysis, ERBB2 amplification was associated with MFS in the whole population and in N- patients. This is consistent with a poor outcome reported in several studies [43, 71‐73].

We found the 20q13 region amplified in 8.5–9.9% of tumors, which is consistent with frequencies previously reported in sporadic breast cancers [3, 27‐29] as well as in familial breast cancer [74]. The 20q13 amplified region was associated with 8q24, 11q13 and 12p13 amplifications. Co-amplification of 11q13/20q13 occurred in 1.6% of cases. The 20q13^Co amplification was associated with high grade and axillary lymph node invasion, in agreement with previous studies [27, 29]. Amplification of 20q13^Z was associated with progesterone receptor negativity and with an accumulation of P53 in cells. Aberrant expression of 20q13 genes may be selected during breast cancer progression because it allows breast cells to overcome senescence and/or apoptosis. This is likely to be due to ZNF217 which promotes immortalization of human mammary epithelial cells [33] and plays a role in suppressing apoptosis [34]. Overexpression of other genes such as AURKA or MYBL2, which encode proteins involved in cell cycle regulation, could generate genomic instability. This accumulation of alterations could then lead to an increased sensitivity to chemotherapy and may explain the better prognosis for patients with amplified 20q13^Co region [29].

We found that the 12p13 amplification was correlated with the basal subtype, which is in agreement with previous expression profiling data [25]. Moreover, the 12p13 amplification was correlated with high grade, absence of steroid receptors and high proliferation index, which are all features of the basal subtype. In contrast, the 11q13 (CCND1) amplification was negatively correlated with the basal subtype and strongly correlated with the luminal A subtype. This is consistent with a previous study [75] that found an inverse correlation between basal-like markers and CCND1 amplification.