Regulation of CD44E by DARPP-32-dependent activation of SRp20 splicing factor in gastric tumorigenesis.

CD44E is a frequently overexpressed variant of CD44 in gastric cancer. Mechanisms that regulate CD44 splicing and expression in gastric cancer remain unknown. Herein, we investigated the role of DARPP-32 (dopamine and cyclic adenosine monophosphate-regulated phosphoprotein, Mr 32000) in promoting tumor growth through regulation of CD44 splicing. Using western blot and quantitative real-time PCR analysis, our results indicated that knockdown of endogenous DARPP-32 markedly reduces the expression of CD44 V8-V10 (CD44E). Using a quantitative splicing luciferase reporter system, we detected a significant increase in the reporter activity following DARPP-32 overexpression (P<0.001). Conversely, knocking down endogenous DARPP-32 significantly attenuated the splicing activity (P<0.001). Further experiments showed that DARPP-32 regulates the expression of SRp20 splicing factor and co-exists with it in the same protein complex. Inhibition of alternative splicing with digitoxin followed by immunoprecipitation and immunoblotting indicated that DARPP-32 has an important role in regulating SRp20 protein stability. The knockdown of endogenous DARPP-32 confirmed that DARPP-32 regulates the SRp20-dependent CD44E splicing. Using tumor xenograft mouse model, knocking down endogenous DARPP-32 markedly reduced SRp20 and CD44E protein levels with a decreased tumor growth. The reconstitution of SRp20 expression in these cells rescued tumor growth. In addition, we also demonstrated frequent co-overexpression and positive correlation of DARPP-32, SRp20 and CD44E expression levels in human gastric primary tumors. Our novel findings establish for the first time the role of DARPP-32 in regulating splicing factors in gastric cancer cells. The DARPP-32-SRp20 axis has a key role in regulating the CD44E splice variant that promotes gastric tumorigenesis.

Characterization of ΔNp73 expression and regulation in gastric and esophageal tumors.

p73 is a member of the p53 protein family. Although the tumor suppressor function of p53 is clearly defined, the role of p73 in tumorigenesis is still a matter of debate. A complex pattern of expression of p73 isoforms makes it difficult to unambiguously interpret the experimental results. Previously, we along with others have found that the N-terminally truncated isoform of p73, ΔNp73, has potent anti-apoptotic and oncogenic properties in vitro and in vivo. In this study, we analyzed, for the first time, the regulation of ΔNp73 in a large number of gastric, gastroesophageal junction and esophageal tumors. We found that expression of ΔNp73 mRNA and protein is increased in these neoplasms. Furthermore, the upregulation of the ΔNp73 protein is significantly associated with poor patient survival. Oncogenic properties of ΔNp73 were further confirmed by finding that ΔNp73 facilitates anchorage-independent growth of gastric epithelial cells in soft agar. As little is currently known about the regulation of ΔNp73 transcription, we investigated the alternative p73 gene promoter that mediates the ΔNp73 expression. Analyzing the ΔNp73 promoter in silico as well as by using chromatin immunoprecipitation, site-directed mutagenesis and deletion analyses, we identified the evolutionary conserved region within the ΔNp73 promoter that contains binding sites for HIC1 (hypermethylated in cancer) protein. We found that HIC1 negatively regulates ΔNp73 transcription in mucosal epithelial cells. This leads to a decrease in ΔNp73 protein levels and may normally control the oncogenic potential of the ΔNp73 isoform.

DNA hypermethylation regulates the expression of members of the Mu-class glutathione S-transferases and glutathione peroxidases in Barrett's adenocarcinoma.

BACKGROUND: The accumulation of reactive oxygen species and subsequent oxidative DNA damage underlie the development of Barrett's oesophagus (BO) and its progression to Barrett's dysplasia (BD) and adenocarcinoma (BAC). METHODS: The promoter regions of 23 genes of the glutathione S-transferase (GST) and glutathione peroxidase (GPX) families were systematically analysed. Quantitative bisulfite pyrosequencing, real-time RT-PCR, western blot and immunohistochemical (IHC) analysis methods were utilised in this study. RESULTS: 14 genes were identified that have CpG islands around their transcription start sites: GSTs (GSTM2-M5, GSTA4, GSTP1, GSTZ1, GSTT2, GSTO1 and GSTO2) and GPXs (GPX1, GPX3, GPX4 and GPX7). Analysis of an initial set of 20 primary samples demonstrated promoter DNA hypermethylation and mRNA downregulation of GPX3, GPX7, GSTM2, GSTM3 and GSTM5 in more than half of the BAC samples. Further analysis of 159 primary human samples (37 normal, 11 BO, 11 BD and 100 BACs) indicated frequent hypermethylation (>or=10% methylation) of GPX3 (62%), GPX7 (67%), GSTM2 (69.1%) and GSTM3 (15%) in BACs. A significant inverse correlation between DNA methylation and mRNA expression level was shown for GPX3 (p<0.001), GPX7 (p = 0.002), GSTM2 (p<0.001) and GSTM5 (p = 0.01). Treatment of oesophageal cancer cell lines with 5-aza-2'-deoxycytidine and trichostatin-A led to reversal of the methylation pattern and re-expression of these genes at the mRNA and protein levels. The IHC analysis of GPX3, GPX7 and GSTM2 on a tissue microarray that contained 75 BACs with normal squamous oesophageal samples demonstrated an absent to weak staining in tumours (52% for GPX3, 57% for GPX7 and 45% for GSTM2) and a moderate to strong immunostaining in normal samples. CONCLUSION: Epigenetic inactivation of members of the glutathione pathway can be an important mechanism in Barrett's tumourigenesis.

The aurora kinase A regulates GSK-3beta in gastric cancer cells.

Aurora kinase A (AURKA) is located at 20q13, a region that is frequently amplified in gastric cancer. In this study, we have investigated the role of AURKA in regulating glycogen synthase kinase (GSK)-3beta and beta-catenin/TCF complex in gastric cancer cells. Our results demonstrate a significant increase in the phosphorylation of GSK-3beta at Ser 9 following the overexpression of AURKA in AGS cells. The immunoprecipitation with antibodies specific for AURKA and GSK-3beta indicated that the two proteins coexist in the same protein complex. The recombinant human AURKA protein phosphorylated the GSK-3beta protein at Ser 9 in a concentration-dependent manner, in vitro. The phosphorylation of beta-catenin (Ser33/37/Thr41) by GSK-3beta is known to target beta-catenin towards degradation. In line with our findings, the increase in phospho-GSK-3beta level was accompanied by a significant decrease in beta-catenin phosphorylation (Ser33/37/Thr41) and accumulation of beta-catenin protein. The knockdown of AURKA reversed the phosphorylation of GSK-3beta and the beta-catenin protein levels. The immunofluorescence analysis demonstrated colocalization of AURKA and GSK-3beta proteins and a significant increase in the nuclear beta-catenin levels in cells overexpressing AURKA. The beta-catenin/TCF transcription activity was measured using the pTopFlash and its mutant pFopFlash luciferase reporter vectors. Indeed, AURKA overexpression led to a significant increase in the pTopFlash reporter activity, whereas kinase dead AURKA mutant (D274A) had no effect. Consistent with these findings, we detected a significant mRNA up-regulation of several direct targets of the beta-catenin/TCF transcription complex (cyclin D1, c-MYC, c-MYC-binding protein, CLDN1, FGF18 and vascular endothelial growth factor), and a two-fold increase in the proliferation rate in AURKA overexpressing cells. We conclude that the AURKA/GSK-3beta interaction is important in regulating beta-catenin, underscoring a novel oncogenic potential for AURKA in gastric tumorigenesis.

Deregulation of MUC4 in gastric adenocarcinoma: potential pathobiological implication in poorly differentiated non-signet ring cell type gastric cancer.

MUC4 is a large, heavily glycosylated transmembrane mucin, that is implicated in the pathogenesis of various types of cancers. To date, no extensive study has been done to check the expression and functional significance of MUC4 in different types of gastric adenocarcinomas. Here, we report the expression profile of MUC4 in gastric adenocarcinomas and its function in poorly differentiated gastric non-signet ring cell carcinoma (non-SRCC) type cells. Immunohistochemical analysis using tissue microarray (TMA) showed a significant difference in MUC4 expression between normal adjacent (n = 45) and gastric adenocarcinoma (n = 83; P < 0.001). MUC4 expression was not associated with tumour type, stage or with the degree of differentiation. To gain further insight into the significance of MUC4 expression in gastric non-SRCC cells, MUC4 was ectopically expressed in AGS, a poorly differentiated gastric non-signet ring cell line. The MUC4 overexpressing cells (AGS-MUC4) showed a significant increase (P < 0.005) in cell motility and a decrease in cellular aggregation as compared with the vector-transfected cells. Furthermore, in vivo tumorigenicity analysis revealed that animals transplanted with the MUC4 overexpressing cells (AGS-MUC4) had a greater incidence of tumours (83%) in comparison to empty vector control (17%). In addition, the expression of MUC4 resulted in enhanced expression of total cellular ErbB2 and phosphorylated ErbB2. In conclusion, our results showed that MUC4 is overexpressed in gastric adenocarcinoma tissues, and that it has a role in promoting aggressive properties in poorly differentiated gastric non-SRCC cells through the activation of the ErbB2 oncoprotein.

DeltaNp73alpha regulates MDR1 expression by inhibiting p53 function.

The p73 protein is a transcription factor and member of the p53 protein family that expresses as a complex variety of isoforms. DeltaNp73alpha is an N-terminally truncated isoform of p73. We found that DeltaNp73 protein is upregulated in human gastric carcinoma suggesting that DeltaNp73 may play an oncogenic role in these tumors. Although it has been shown that DeltaNp73alpha inhibits apoptosis and counteracts the effect of chemotherapeutic drugs, the underlying mechanism by which this p73 isoform contributes to chemotherapeutic drug response remains to be explored. We found that DeltaNp73alpha upregulates MDR1 mRNA and p-glycoprotein (p-gp), which is involved in chemotherapeutic drug transport. This p-gp upregulation was accompanied by increased p-gp functional activity in gastric cancer cells. Our data suggest that upregulation of MDR1 by DeltaNp73alpha is mediated by interaction with p53 at the MDR1 promoter.

A novel method for gene expression mapping of metastatic competence in human bladder cancer.

Expression profiling by DNA microarray analysis has provided insights into molecular alterations that underpin cancer progression and metastasis. Although differential expression of microarray-defined probes can be related to numerical or structural chromosomal alterations, it is unclear if such changes are also clustered in distinct chromosomes or genomic regions and whether chromosomal alterations always reflect changes in gene expression. Here we apply the dChip algorithm and a novel technique to test the hypothesis that expression changes occurring as a function of tumor progression and metastasis are nonrandomly distributed. Expression profiling of a human xenograft model of lung metastasis phenotype indicates that chromosomes 2, 11, and 20 contain higher percentages of differentially expressed genes (P < .05). Furthermore, we show that a number of differentially expressed probes mapped to chromosome 17q, defining the existence of an expression "hot spot" corresponding to an area of gain determined by comparative genomic hybridization (CGH). Interestingly, other areas of gains detected by CGH were not associated with expression hot spots. In summary, we show that gene expression changes during bladder cancer lung metastasis occur nonrandomly in specific chromosomes and intrachromosomal locations.

The gastrin gene promoter is regulated by p73 isoforms in tumor cells.

p73, a new p53 family member, is a transcription factor that is increasingly recognized in cancer research as an important player in tumorigenesis as well as in chemotherapeutic drug sensitivity. Despite the substantial structural and functional similarities to p53, accumulating evidence suggests that p53 and p73 may differently regulate their transcriptional targets. In this study, we have investigated the role of p73 in regulation of the gastrin gene promoter. Gastrin is a peptide hormone and an important factor in determining the progression of a number of human malignancies. Our results show that p73 can bind to the gastrin promoter. This leads to transcriptional upregulation of gastrin mRNA. We also found that the levels of gastrin and p73 transcripts correlate in primary gastric tumors. Taken together, our results demonstrate a novel mechanism for regulation of gastrin gene transcription and support a concept that p53 and p73 may have different biological roles in tumors.

Molecular targets for tumour progression in gastrointestinal stromal tumours.

BACKGROUND AND AIMS: The distinction between benign and malignant gastrointestinal stromal tumours (GISTs) is often unclear at the clinical and histopathology levels. GISTs are believed to arise from the stem cells of Cajal. In order to define genetic biomarkers and identify target genes related to GIST progression, we analysed and compared benign and malignant GISTs with verified follow up data using cDNA expression arrays. METHODS: Eight genes were frequently overexpressed in malignant GISTs and their overexpression was confirmed using quantitative real time reverse transcription-polymerase chain reaction. These genes included ezrin (villin 2 (VIL2)), collagen 8 alpha 1 subunit (COL8A1), G2/mitotic specific cyclin B1 (CCNB1), high mobility group protein (HMG2), TSG101 tumour susceptibility protein, CENP-F kinetochore protein, protein tyrosine kinase 2 (FAK), and protein kinase DYRK2. To test these genes in a clinical setting, we obtained diagnostic samples of 16 additional GISTs that were classified at diagnosis as benign, malignant, and uncertain malignant potential (UMP). RESULTS: There was remarkable gene overexpression in all malignant GISTs. Statistical analyses revealed significant correlations between overexpression of several gene pairs in malignant GISTs. We found the strongest correlations (rho>0.70) among the significant correlations (p<0.01) between CCNB1-CENP-F (rho = 0.87) and CCNB1-FAK (rho = 0.73). Gene expression of the UMP GISTs suggested two different groups. Three UMP GISTs had gene expression consistent with malignant tumours and their follow up data revealed that indeed these patients had recurrences later on. On the other hand, UMP GISTs that had low gene expression levels continued free of disease for several years. CONCLUSIONS: These results provide insight into the oncogenesis of GISTs and suggest that testing the expression profile of a number of genes may segregate GISTs into groups of different tumour behaviour.

Inactivation of the E-cadherin gene in sporadic diffuse-type gastric cancer.

Loss of the cell adhesion molecule E-cadherin has been observed in a variety of human carcinomas, and germline E-cadherin mutations have been found in several familial cases of diffuse gastric cancer. We sought to determine the prevalence and nature of E-cadherin alterations in "sporadic" gastric carcinomas. We performed comprehensive sequencing of the coding region, loss of heterozygosity (LOH) analysis, and immunohistochemical protein expression determination on 40 sporadic gastric adenocarcinomas. In total, 7 of 25 diffuse-type cancers harbored genetic alterations in the E-cadherin gene. Novel mutations predicted to significantly compromise protein function were found within 4 of these cancers, 2 of which harbored alterations resulting in biallelic inactivation of the gene product. Three diffuse cancers failed to amplify Exon 8 of E-cadherin, suggesting the presence of a homozygous abnormality. Notably, one germline E-cadherin mutation was also identified within these "sporadic" diffuse cancers. Significant gene mutations were not found in the 14 intestinal-type or histologically mixed cancer. Immunohistochemistry revealed aberrant or negative protein expression in seven diffuse-type tumors, four of which correlated with the genetic alterations. Both diffuse and intestinal-type tumors exhibited low rates of LOH, suggesting that allelic loss at the locus is not a common mechanism for E-cadherin inactivation during gastric tumorigenesis. Our observations suggest that inactivation of the E-cadherin gene occurs only in a subset of diffuse-type gastric cancers, as the majority of cases did not contain genetic alterations or identifiable protein abnormalities. Germline E-cadherin alterations, although rare, may underlie some diffuse gastric cancer cases that have important biologic and practical implications

Genetic differences between adenocarcinomas arising in Barrett's esophagus and gastric mucosa.

BACKGROUND & AIMS: Barrett adenocarcinoma (BA+) and gastric adenocarcinoma comprise a related group of neoplasms that nevertheless have some distinct clinicopathologic characteristics. This study aimed at defining critical molecular abnormalities that may underlie differences between BA+ and gastric adenocarcinomas. METHODS: We used comparative genomic hybridization for the analyses of 34 xenografts of adenocarcinomas that arose from esophageal or gastric origin. RESULTS: All tumors, except one, exhibited DNA copy number alterations. Losses in 4q and 14q and gains at 2p and 17q were more frequent in proximal (esophageal, gastroesophageal junction [GEJ], and cardia) tumors than in distal (body and antrum) tumors (P

Novel DNA copy number losses in chromosome 12q12--q13 in adenoid cystic carcinoma.

In order to find common genetic abnormalities that may identify loci of genes involved in the development of adenoid cystic carcinoma (ACC), we investigated DNA copy number changes in 24 of these tumors by comparative genomic hybridization (CGH). Our results indicate that unlike many carcinomas, ACCs have relatively few changes in DNA copy number overall. Twenty tumors had DNA copy number changes, which were mostly restricted to a few chromosomal arms. A frequent novel finding was the loss of DNA copy number in chromosome 12q (eight tumors, 33%) with the minimal common overlapping region at 12q12--q13. Deletion in this region has not been reported to be frequent in other types of cancer analyzed by CGH. In addition, deletions in 6q23-qter and 13q21--q22 and gains of chromosome 19 were observed in 25% to 38% of ACCs. Deletion of 19q, previously reported in a small series of ACC, was not identified in the current group of carcinomas. The current CGH results for chromosomes 12 and 19 were confirmed by microsatellite allelotyping. These results indicate that DNA copy number losses in 12q may be important in the oncogenesis of ACC and suggest that the 12q12--q13 region may harbor a new tumor-suppressor gene.

Concomitant DNA copy number amplification at 17q and 22q in dermatofibrosarcoma protuberans.

Dermatofibrosarcoma protuberans (DFSP) is a tumor of low or intermediate malignant potential with a tendency for recurrence, but low rate of metastasis. The tumorigenesis of DFSP has recently been shown to be associated with the fusion of the collagen type I alpha 1 (COL1A1) and platelet-derived growth factor B-chain (PDGFB) genes, often as a consequence of translocation t(17;22)(q22;q13). Cytogenetically, DFSP is often characterized by supernumerary ring chromosomes containing material from chromosomes 17 and 22. A subset of DFSPs undergo fibrosarcomatous transformation de novo or upon recurrence, and contain components indistinguishable from fibrosarcoma (FS-DFSP). The fibrosarcomatous transformation appears to carry an increased risk for recurrence and metastasis, and is considered to represent tumor progression. The molecular cytogenetic events contributing to tumor progression are unknown. We used comparative genomic hybridization to analyze DNA copy number changes in 11 cases of typical DFSP and 10 cases of FS-DFSP. All cases in both groups were found to exhibit a gain or high-level amplification on chromosome 17q and the majority also on 22q. This finding is in line with previous studies, and suggests further that not only the COL1A1/PDGFB fusion gene formation but also the role of DNA copy number gains in the 17q and 22q regions is crucial per se in the pathogenesis of DFSP. Even though FS-DFSPs displayed a trend toward increase in the number of DNA copy number changes, the difference was not statistically significant, which indicates that mechanisms other than copy number changes are important in the transformation process of DFSP.

Changes in gene expression during progression of ovarian carcinoma.

The molecular events leading to the development and progression of serous ovarian carcinoma are not completely understood. We performed a large scale survey for the identification of differentially expressed genes in serous ovarian carcinoma by using cDNA array analysis. Differences in gene expression between serous adenocarcinoma and benign serous adenoma, and between advanced and/or moderately or poorly differentiated and local, highly differentiated serous adenocarcinoma were assessed. The most striking difference between adenocarcinoma and benign adenoma was upregulation of RHOGDI2 in the carcinomas irrespective of the clinical tumor stage. Other changes in carcinoma were upregulation of MET and Ne-dlg, and downregulation of HGFAC, desmin, and PDGFA. The most prominent differences between advanced and local adenocarcinoma were upregulation of COL3A1, CFGR, and MET in advanced carcinoma, and downregulation of HGFAC, FZD3, and BFL1 in the same tumors. In conclusion, significant differences were found in the gene expression between benign and malignant serous ovarian tumors, and between local, highly differentiated and advanced and/or moderately or poorly differentiated serous adenocarcinomas. The differentially expressed genes may be related to the carcinogenesis and progression of the malignant growth.

Expression profiling of gastric adenocarcinoma using cDNA array.

To investigate the expression profile of gastric adenocarcinoma, cDNA array experiments were performed using Atlas Human Cancer 1.2 K Array (Clontech Laboratories, Palo Alto, CA) on nine xenografted and two primary gastric cancer samples. The expression of the tumor samples was compared to that of two normal gastric epithelial tissues. The expression pattern of the primary tumors was similar to that of xenografted tumors. The up-regulated genes had expression ratios ranging from 2.5 to 16, whereas the down-regulated genes had a range from -2.5 to -16. No variation in gene expression was detected in the analysis of the xenografted tumors versus the primary tumors, indicating that the xenografts represented the primary tumors well. Thirty-eight genes showed altered gene expression in 5 or more samples (>45%). Thirty-one genes were up-regulated and seven genes were down-regulated. The most abundantly up-regulated genes (ratio >5) included genes such as S100A4, CDK4, MMP14 and beta catenin. The GIF was markedly down-regulated (ratio < -10). To confirm our findings, six genes (three up- and three down-regulated) were selected for semi-quantitative RT-PCR analysis. The RT-PCR results were consistent with the array findings. Our approach revealed that several genes are abnormally expressed in gastric cancer and found that genes known to interact in several common molecular pathway(s) were consistently altered.

Gene expression profiling of malignant mesothelioma cell lines: cDNA array study.

To reveal genes relevant for malignant mesothelioma (MM), we carried out cDNA array experiments on 4 MM cell lines and 2 primary mesothelial cell cultures established from pleural fluid of non-cancer patients. Human cancer gene filters including 588 genes were used for the cDNA array experiments. Our study revealed 26 over-expressed genes that play a role in the regulation of cell fate, cell cycle, cell growth and DNA damage repair and 13 under-expressed genes encoding growth factors, receptors and proteins involved in cell adhesion, motility and invasion to be common to 3 or 4 MM cell lines. We confirmed the cDNA array results using RT-PCR for 5 of the over-expressed genes and for 3 of the under-expressed genes. Our study presents gene expression profiles in MM cell lines and shows the involvement of several genes, such as those encoding JAGGED1, ser/thr protein kinase NIK, Ku80 and cyclin D2, novel in MM.

Comparative genomic hybridization technique.

Screening for chromosomal changes in solid tumors was long hindered by methodological problems encountered in standard cytogenetic analysis. Comparative genomic hybridization (CGH), a technique that emerged in 1992 (1) has proved to be a powerful tool for molecular cytogenetic analysis of neoplasms. The main prerequisite of the technique is DNA isolated from tumor samples. As no cell culture of tumor material is required, the technique has been successfully used to study fresh and frozen tissue samples, as well as archival formalin-fixed paraffin-embedded tissue samples. CGH allows to screen entire tumor genomes for gains and losses of DNA copy number, enabling consequent mapping of aberrations to chromosomal subregions. The technique is based on fluorescence in situ hybridization. Tumor and reference DNA are differentially labeled with fluorochromes (green and red, respectively) and mixed in equal amounts. The mixture is cohybridized competitively to a normal metaphase slide prepared from a lymphocyte cell culture of a normal healthy individual. After hybridization and washes, the chromosomes are counterstained with DAPI (blue) and slides are mounted with an antifading medium.

Genetic analysis using microarrays.

A vast amount of genome sequencing data has become available over the past few years and methods to facilitate high-throughput analysis of large sets of genes and samples have been developed to localize novel genes related to human cancer. As advanced robotic applications have made it possible to manufacture high-precision microarrays on glass or membranes, pioneering scientists have introduced several variants of the "array" technology: oligonucleotide arrays (1), DNA microarrays (CGH arrays) (2), tissue microarrays (3), and cDNA microarrays (4,5). The array technology is based on fluorescently (glass-based arrays, chips) or radioactively (filter-based array) labeled nucleic acids that are hybridized to the microarray and imaged with a laser scanner or a phosphor imager, respectively. The images are then processed using microarray analysis software. These techniques have recently been reviewed in detail in Nature Genetics (vol. 21, Suppl. 1, 1999).

DNA sequence copy number changes in gastrointestinal stromal tumors: tumor progression and prognostic significance.

To identify genetic changes related to tumor progression and find out diagnostic and prognostic genetic markers in gastrointestinal stromal tumors (GISTs), 95 tumor samples (24 benign GISTs, 36 malignant primary GISTs, and 35 GIST-metastases) from 60 patients were studied using comparative genomic hybridization. DNA copy number changes were detected in all samples. Benign GISTs had a mean of 2.6 aberrations/ sample (losses:gains, 5:1) and significantly fewer DNA copy number changes and fewer gains than malignant primary and metastatic GISTs (P < 0.01). High-level amplifications were not seen in benign GISTs. Malignant primary GISTs had a mean of 7.5 aberrations/tumor (losses: gains, 1.6:1), whereas the mean number of aberrations/metastatic GIST was 9 (losses:gains, 1.8:1). Frequent changes observed in all GIST groups included losses in chromosome arms 1p (51%), 14q (74%), and 22q (53%). Gains and high-level amplifications at 8q and 17q were significantly more frequent in metastatic GISTs (57 and 43%) than in benign GISTs (8 and 0%; P < 0.001) and malignant primary GISTs (33 and 25%; P < 0.05). Gains and high-level amplifications at 20q were only seen in malignant primary and metastatic GISTs (P < 0.01), and gains at 5p were not detected in benign GISTs (P < 0.01). Losses in chromosome arm 9p were never seen in benign tumors (P < 0.001), and they were more frequent in metastatic GISTs than in malignant primary GISTs (63 and 36%; P < 0.05). Losses in 13q were less frequent in benign GISTs than in malignant primary (P < 0.05) and metastatic (P < 0.01) GISTs. Our results show that several DNA copy number changes are related to the behavior of GISTs and can be used as prognostic markers for tumor progression.