Genomic imbalances in esophageal squamous cell carcinoma identified by molecular cytogenetic techniques

This review summarizes the chromosomal changes detected by molecular cytogenetic approaches in esophageal squamous cell carcinoma (ESCC), the ninth most common malignancy in the world. Whole genome analyses of ESCC cell lines and tumors indicated that the most frequent genomic gains occurred at 1, 2q, 3q, 5p, 6p, 7, 8q, 9q, 11q, 12p, 14q, 15q, 16, 17, 18p, 19q, 20q, 22q and X, with focal amplifications at 1q32, 2p16-22, 3q25-28, 5p13-15.3, 7p12-22, 7q21-22, 8q23-24.2, 9q34, 10q21, 11p11.2, 11q13, 13q32, 14q13-14, 14q21, 14q31-32, 15q22-26, 17p11.2, 18p11.2-11.3 and 20p11.2. Recurrent losses involved 3p, 4, 5q, 6q, 7q, 8p, 9, 10p, 12p, 13, 14p, 15p, 18, 19p, 20, 22, Xp and Y. Gains at 5p and 7q, and deletions at 4p, 9p, and 11q were significant prognostic factors for patients with ESCC. Gains at 6p and 20p, and losses at 10p and 10q were the most significant imbalances, both in primary carcinoma and in metastases, which suggested that these regions may harbor oncogenes and tumor suppressor genes. Gains at 12p and losses at 3p may be associated with poor relapse-free survival. The clinical applicability of these changes as markers for the diagnosis and prognosis of ESCC, or as molecular targets for personalized therapy should be evaluated.


Esophageal Carcinoma
Cancer of the esophagus is the ninth most common malignancy in the world, but its incidence varies considerably among geographical regions (Lam, 2000), with a high incidence in China, Japan, Singapore and Puerto Rico (INCA, 2009a). The American Cancer Society estimated that around 16,470 new cases of esophageal carcinoma occurred in the USA in 2008 (American Cancer Society, 2009). In Brazil, esophageal cancer ranked sixth in cancer mortality in 2000, with 5,307 deaths; about 10,550 new cases were reported in 2008, with an incidence of 1.04-19.07 per 1000,000 males and 0.39-7.58 per 100,000 females (INCA, 2009b).
The development of human esophageal cancer is progressive, involving the accumulation of genetic changes that culminate in malignant transformation (Knudson, 1985;Somers and Schechter, 1992;Xue et al., 2006). An early indicator of this process is the increased proliferation of esophageal epithelial cells that morphologically pro-gresses to basal cell hyperplasia, dysplasia, carcinoma in situ and invasive carcinoma (Muñoz, 1997;Mandard et al., 2000;D'Amico, 2006).
With the exception of the USA, esophageal squamous cell carcinoma (ESCC) is histologically the most prevalent type of esophageal cancer worldwide and has a multifactorial origin. In addition to environmental components (Crawford, 2004), several genetic factors are associated with esophageal carcinogenesis, such as chromosomal aneuploidy, allelic deletions, activation of oncogenes and inactivation of tumor suppressor genes (Kuwano et al., 2005). At the cellular level, these factors lead to disorders of cell proliferation, differentiation and apoptosis (Koch et al., 1994;McCabe and Dlamini, 2005;D'Amico, 2006;Daigo and Nakamura, 2008;Khushalani, 2008).
Specific chromosomal aberrations have been identified as markers for the diagnosis and prognosis of solid tumors (Tada et al., 2000;Yen et al., 2001;Shiomi et al., 2003;Qin et al., 2004Qin et al., , 2005aQin et al., ,b, 2008Wang et al., 2006). Despite reports of numerous chromosomal alterations, no particular diagnostic or prognostic chromosomal markers have been described for esophageal carcinoma. The aim of this review is to summarize the recurrent chromosomal changes, such as gains and losses in regions that may harbor oncogenes and tumor suppressor genes that have been detected in ESCC by molecular cytogenetic approaches. These chromosomal imbalances may represent clinically relevant markers for the diagnosis and prognosis of ESCC or for the identification of novel therapeutic targets.

Molecular Cytogenetic Technologies Identify Genomic Changes In Cancer
The primary cytogenetic technique used to investigate the molecular pathogenesis of esophageal carcinogenesis is DNA fluorescence in situ hybridization (FISH). This technique uses small fragments of DNA as fluorescent probes that bind to specific chromosomal sequences of the target DNA to which they show a high degree of complementarity (Bauman et al., 1980;Langer et al., 1981). FISH probes are often derived from DNA fragments that vary from a few hundred to 200,000 base-pairs; these fragments are isolated, purified, amplified and labeled with fluorochrome-conjugated nucleotides. The DNA probes hybridize to distinct DNA targets such as metaphase chromosomes, interphase nuclei, and extended chromatin fibers and DNA fragments in a variety of biological specimens or platforms, including isolated cells, tissue sections and bacterial artificial chromosomes (BAC) and oligonucleotide arrays (Solinas-Toldo et al., 1997;Pinkel et al., 1998;Speicher and Carter, 2005).
There are numerous variants of the FISH assay, with the most effective for detecting extensive genomic imbalances being comparative genomic hybridization (CGH). CGH was originally developed as the competitive hybridization of a mixture of test DNA and normal reference DNA, labeled with different fluorochromes, to metaphase spreads of a normal specimen, i.e., metaphase CGH (mCGH) (Kallioniemi et al., 1992;du Manoir et al., 1993). In these conditions, chromosomal regions of the test DNA that have normal copy numbers will show a balanced ratio of hybridization with the test and control DNA, whereas chromosomal regions with an excess or loss of copy numbers in the test DNA will show predominantly the color of the hybridization in the normal template. The test DNA used in the CGH assay can be extracted from dividing or non-dividing cells from virtually all types of tissues, including formalin-fixed specimens; the latter possibility allows an informative overview of archived specimens. The level of resolution of specific imbalances in the mCGH assay depends essentially on the condensation of the chromosomes to which the DNA mix is hybridized and ranges from 5 to 10 Mb (Kallioniemi et al., 1992;Speicher and Carter, 2005). Although it is possible to identify large regions involved in low level of genomic gain or loss and small regions with focal amplification, the resolution of the mCGH is generally limited or insufficient for the identification of specific chromosomal bands (Kallioniemi et al., 1992;Speicher and Carter, 2005). Consequently, mCGH is gener-ally used as a preliminary tool to infer potential genes located in imbalanced regions and to confirm or support the results of studies with higher resolution techniques, such as array CGH (aCGH) or FISH with single gene probes (Nakakuki et al., 2002;Arai et al., 2003) The aCGH technique detects changes in chromosome copy number at a much higher resolution than mCGH (Solinas-Toldo et al., 1997;Pinkel et al., 1998). Instead of using metaphase spreads as the template for hybridization, the aCGH uses a collection of DNA inserts contained in bacterial artificial chromosomes (BAC arrays) or oligonucleotides (oligo-arrays) printed on a glass slide. As with the mCGH, a differentially labeled mixture of DNAs from the test sample and a normal reference control sample is hybridized with the selected platform and the ratio of the fluorescence intensity of the test to reference DNA is calculated. Using aCGH, changes in copy number can be detected based on a few hundred kilobases of DNA sequences for BAC arrays or 30 kb for oligoarrays (Snijders et al., 2001;Fiegler et al., 2003;Medical Genetics Laboratories, 2008). More recently, imbalances involving single nucleotides have been detected by single nucleotide polymorphism (SNP)-based arrays, a type of DNA microarray used to detect polymorphisms within a population (Sherry et al., 2001).
Other variants of FISH technology include multiplex-FISH (M-FISH) and spectral karyotyping (SKY). These procedures allow identification of the origin of each chromosomal region in a metaphase cell by visualizing all 24 human chromosomes in a single hybridization (Speicher and Carter, 2005). In both techniques, the probe set is a pool of differentially labeled DNAs for each of the 24 human chromosomes. In M-FISH, the images for each fluorochrome are collected individually and merged, and a combinatorial labeling algorithm identifies each chromosome that is then visualized in pre-defined pseudocolor (Speicher et al., 1996). In the SKY assay, a single image is captured per cell and an interferometer is used to discriminate the fluorochrome spectrum in each pixel, to which a pseudocolor is assigned (Schröck et al., 1996). Both techniques have been successful in clarifying complex chromosomal rearrangements in solid tumors (Schröck and Padilla-Nash, 2000), including ESCC (Yen et al., 2003).
Together, the numerous variants of FISH technology have allowed the accurate identification of chromosomal DNA sequences of interest and facilitated the screening of the whole genome for gains and losses associated with carcinogenesis.

Genomic Imbalances In Escc Cell Lines
The few molecular cytogenetic studies of ESCC cell lines reported to date indicate that these cell lines are highly abnormal cytogenetically. In a study in which eight ESCC cell lines were screened using mCGH, SKY and FISH with single probes, the pooled CGH results revealed frequent gains in almost all chromosome arms (1p,1q,3q,5p,6p,7p,7q,8q,9q,11q,12p,14q,15q,16p,16q,17q,18p,19q,20q,22q and Xq), with frequent losses on 3p, 4, 5q, 6q, 7q, 9p and 18q. SKY analyses detected 195 translocations, 13 deletions and two duplications in the eight cell lines, with the most frequently amplified genes being PIK3CA (3q26) and TP63 (3q28). PIK3CA encodes the catalytic subunit of phosphatidylinositol 3-kinase (PIK3), which uses ATP to phosphorylate phosphatidylinositol (National Center for Biotechnology Information, 2009a) and TP63 encodes a protein involved in the development and maintenance of stratified epithelial tissues (National Center for Biotechnology Information, 2009b). These oncogenes were amplified in six and five cell lines, respectively (Yen et al., 2003).
Based on previous CGH studies that showed frequent amplifications in 18p in esophageal cell lines (Shimada et al., 1992;Pimkhaokham et al., 2000), Nakakuki et al. (2002) used FISH to screen 29 ESCC cell lines and identified amplifications of 14 known genes and 21 uncharacterized transcripts in chromosome 18 amplicons. These authors also investigated the corresponding levels of gene expression by Southern-, dot-and northern-blotting. Only four known genes (YES1, TYMS, HEC and TGIF) showed amplification and corresponding over-expression. YES1 encodes a protein with tyrosine kinase activity, TYMS is critical for DNA replication and repair, HEC is involved in spindle checkpoint signaling and TGIF is a highly con-served transcription regulator with a potential role in the transmission of nuclear signals during development and in adults. These findings suggested that these genes are involved in 18p11.3 amplification and may be associated with esophageal tumorigenesis. Figure 1A summarizes these studies. The alterations detected affected most of the genome and involved regions harboring many known oncogenes and tumor suppressor genes, as well regions not yet associated with such genes. Although the level of molecular resolution of most of these studies is low and inconclusive, these findings are promis-Esophagus cancer: unbalanced genes 207  2q, 3q, 5p, 6p, 7, 8q, 9q, 11q, 12p, 14, 15, 16, 17, 18p, 19q, 20q, 22q 4,5q,6q,7q,8p,9,10p,12p,13,14p,15p,18,20,22,Xp and Y. (B) In clinical specimens, chromosomal gains were common in 1q, 2q,3q,5p,6,7,8q,9,10,11q,12,13q,14q,16p,17,18q,20 and Xq,specifically  ing in that they provide a starting point for further investigations on the molecular pathogenesis of ESCC and the development of new therapeutic approaches for such cancer.

Genomic Imbalances In Escc Tumors
Despite advances in our understanding of the risk factors and cellular derangements associated with esophageal cancer, the clinical treatment of this disease remains largely unaltered and long-term survival from this cancer remains poor, with a 5-year survival rate of~20% .

Early Genomic Changes and Imbalances Associated With Tumor Staging
ESCC arises through multi-step genetic and cytogenetic alterations. However, the time sequence of these al-terations remains to be determined. In this regard, studies in which chromosomal aberrations are correlated with the stage and clinical outcome of prognostic significance are necessary in order to facilitate the selection of patients for specific treatments.
An interesting recent study explored the usefulness of M-FISH for the early diagnosis and risk prediction of precursor lesions of ESCC in tumor and premalignant lesions in 113 patients (Yao et al., 2008). Elevated rates of aneuploidy were frequently observed in chromosomes 3, 8, 10, 12, 17 and 20 in ESCC and its precursor dysplastic lesions. These findings support the conclusion that the application of a multi-target FISH assay to investigate chromosomal aneuploidy at esophageal dysplastic sites may be useful in predicting the risk of ESCC.
Gains involving 3q, 5p, 1q and 11q13-14 and losses involving 4 and 13q are significantly correlated with the pathological stage, whereas a gain of 8q and loss of 4p are linked to nodal metastasis; similarly, a gain of 2p and loss of 4 and 11q14-qter are associated with distant organ metastasis (Qin et al., 2004). Gains involving 1q, 3q, 5p and 11q13-14 and losses involving 4 and 13q are critical for the development of ESCC, whereas a gain of 2p and 8q and loss of 4 and 11q14-qter are later events associated with tumor progression and thought to confer metastatic potential to the disease. Nodal and distant organ metastases apparently involve different genes (Qin et al., 2005b). Gains involving 3q and 11q13 and losses involving 3p, 4q, 5q14-23, 9p and 18q have been detected in early and advanced stages of ESCC.
Deletions of 4p and 13q12-q14 and a gain of 5p are significantly correlated with the pathological state. Losses of 8p22-pter and 9p are more frequent in patients with advanced disease. A gain of 8q24-qter is more frequent in patients with grade 3 tumors (Yen et al., 2001). Using mCGH, Shiomi et al. (2003) observed that gains involving 3q, 8q, 11q13 and 14q were early events, while the loss of 3p, 5q, 13q and 21q and gain of 1p and Xq were later events in the development of individual tumors.
Gains of 6p and 20p and losses of 10p and 10q are the most significant imbalances in primary carcinoma and metastasis, which suggests that these regions may harbor oncogenes and tumor suppressor genes (Qin et al., 2005a).
A gain of 12p and loss of 3p has been associated with poor relapse-free survival .

The Search For Relevant Oncogenes In Escc
In 41 primary ESCC investigated with CGH (Fujita et al. 2003), the expression of numerous genes, including the cell cycle-regulator kinase gene BTAK (20q13.2-3) and E2F1 (20q11.2), which plays a crucial role in cell cycle regulation, was enhanced in~10% of tumors. Other genes, such as NCOA3 (20q12), which encodes a nuclear receptor co-activator that enhances the transcriptional activator function of nuclear hormone receptors, and DcR3 (20q13.3), which regulates apoptosis were up-regulated to a lesser extent. Xu et al. (2007) detected NCOA3 overexpression and an increased copy number in 46% and 13% of 221 ESCCs, respectively. NCOA3 overexpression was observed more frequently in late compared to early stages, but there was no significant association between the expression of NCOA3 and lymph node metastases. These observations suggest that overexpression of NCOA3 as a result of genomic gain or other molecular mechanisms might provide a selective advantage for the development and local invasion of certain subsets of ESCC.
To date, there has been no detailed analysis of loss of heterozygosity (LOH) for chromosome 18q in ESCC. However, LOH on chromosome 18q is common in several cancers, with frequencies of 55%-67% in colorectal cancer (Jen et al., 1994;Thiagalingam et al., 1996), 90% in pancreatic cancer , 59% in ovarian cancer (Lassus et al., 2001) and 40%-84% in head and neck squamous cell carcinoma (Papadimitrakopoulou et al., 1998;Pearlstein et al., 1998;Takebayashi et al., 2000). In these cancers, frequent LOH on chromosome 18q correlates with tumor growth, aggressive tumor behavior and tumorigenesis. These findings suggest that chromosome 18q may harbor tumor suppressor genes for various cancers (Ando et al., 2007). In a FISH analysis of the allelic imbalance of chromosome 18q in ESCC resected samples from two out of five patients showed a loss of one copy of chromosome 18q, and 13 of 46 ESCC samples (28.3%) showed loss of almost all of chromosome 18q (Ando et al., 2007). The authors suggested that the loss of 18q may play an important role in the progression of ESCC.
Few studies have focused on the identification of altered pathways and clinically applicable markers. Oncogene amplification was examined by DNA microarrays in 20 surgically resected ESCC and 57 oncogenes were found to be amplified. Alterations in DNA copy number detected by microarrays were compared to those obtained by mCGH. DNA microarrays showed that eight oncogenes (CCND1,FGF3/ FGF4,EMS1,SAS,ERBB2,PDGFRA,MYC and BCL2) were amplified in 9 of 20 tumors. Although ERBB2 expression was 23-fold greater than the basal level in one case, the average level of gene amplification was generally only 2-4-fold above the control value.
EMS1, CCND1 and FGF3/FGF4, which are all located on 11q13, were amplified in 7, 5 and 4 of 20 ESCC, respectively, and were co-amplified in three tumors. EMS1 regulates the interactions between components of adherens-type junctions and organizes the cytoskeleton and cell adhesion structures of epithelia and carcinoma cells (National Center for Biotechnology Information, 2009e). FGF3/FGF4 broadly regulate mitogenesis, cell survival and oncogenic activity.
A comparison of genomic DNA microarray and mCGH data showed that although most of the amplified oncogenes were included in chromosomal regions for which gains in DNA copy number were detected by mCGH, not all of the amplified genes detected by microarrays showed concomitant gains in the DNA copy number in mCGH; this lack of correlation between the two techniques confirmed the limited resolution of the mCGH assay. Microarrays of oncogenes are useful for the comprehensive identification of amplified oncogenes and for analyzing specific chromosomal regions in which mCGH analysis indicates an increase in DNA copy number (Arai et al., 2003).
FISH assays revealed the amplification of PLK1 (polo-like kinase 1), an essential gene for the maintenance of genomic stability during mitosis (Feng et al., 2009); this gene may therefore be a useful prognostic marker. An analysis of 108 ESCCs and nine ESCC cell lines revealed a frequent gain of the genes MDS1 (myelodysplasia syndrome 1) and PRKCI (protein kinase C, iota), which have been implicated in neoplastic transformation, and a positive correlation between the level of PRKCI expression and tumor size, lymph node metastasis and clinical stage. PRKCI gene amplification was highly correlated with protein overexpression (Yang et al., 2008b).
EGFR expression is enhanced in many cancers and is sometimes accompanied by gene amplification. FISH analysis has shown that the EGFR gene is amplified in ESCC (Sunpaweravong et al., 2005;Hanawa et al., 2006). Indeed, EGFR gene overrepresentation (balanced gene and chromosome 7 polysomy) and HER-2 amplification are common events in ESCC (Mimura et al., 2005;Sunpaweravong et al., 2005;Bizari et al., 2006).
Cyclin D1 (CCDN1) is a cell-cycle regulator and oncogene implicated in the pathogenesis of numerous types of tumors. Amplification of the CCDN1 gene is common in ESCC and has been detected by FISH (Sheyn et al., 1997;Jin et al., 2004;Manoel-Caetano et al., 2004;Sunpaweravong et al., 2005;Bizari et al., 2006).
The correlation between the overexpression and amplification of oncogenes and patient survival has not been extensively investigated. However, the survival of patients with increased expression of BTAK or E2F1 is significantly lower than that of patients without this alteration, which suggests that overexpression of these genes probably increases the number of malignant ESCC phenotypes and Esophagus cancer: unbalanced genes 209 may be a useful marker of poor prognosis (Fujita et al., 2003).

Concluding Remarks
Various molecular cytogenetic approaches have been used to demonstrate the extensive genetic complexity associated with the different stages of ESCC and have identified key genes involved in esophageal cancer. Most of the unbalanced chromosomal regions are similar in primary tumors and cell lines, thus confirming that cell lines provide a reliable model for investigating the molecular mechanisms involved in ESCC.
Specific chromosomal imbalances are associated with the progression of esophageal tumors. Whereas gains in 1q, 5p, 8q, 14q and losses in 4p, 13q and 18q are associated with the early stages of ESCC development, gains in 1p, 2p, 7p22.3, 8q, 8q24-qter and Xq and losses in 8p22pter, 11q14-qter, 13q and 21q are associated with advanced stages. Gains involving 3q and 11q13 and losses involving 3p, 4q, 5q14-23, 9p and 18q are detected in both early and advanced stages of ESCC. These regions harbor genes associated with essential cellular processes, such as signal transduction, transcriptional regulation, cell proliferation and cell differentiation, involved in signalizing cascades related to cancer. Some imbalances, such as losses involving 4p, 9p and 11q and gains involving 1p36-32, 5p, 7q and 19p13.3, are poor prognostic markers. Gene overexpression most frequently affects oncogenes such as YES1, TYMS, HEC, TGIF, NCOA3, BTAK, DCR3, E2F1, MYC, EGFR, EGR2, CCND, FGF3/FGF4, EMS1, SAS, ERBB2, PDGFR1, BCL2, MDS1 and PRKCI, whereas genomic losses lead to the deletion of suppressor tumor genes such as CDKN2A, MTAP and TP53. Identification of the role of these critical genes in esophageal carcinogenesis will assist the development of individualized target therapy that should in turn improve the clinical outcome of ESCC patients.