Matrix-based comparative genomic hybridization: Biochips to screen for genomic imbalances

Comparative genomic hybridization (CGH) to metaphase chromosomes has been widely used for the genome-wide screening of genomic imbalances in tumor cells. Substitution of the chromosome targets by a matrix consisting of an ordered set of defined nucleic acid target sequences would greatly enhance the resolution and simplify the analysis procedure, both of which are prerequisites for a broad application of CGH as a diagnostic tool. However, hybridization of whole genomic human DNA to immobilized single-copy DNA fragments with complexities below the megabase pair level has been hampered by the low probability of specific binding because of the high probe complexity. We developed a protocol that allows CGH to chips consisting of glass slides with immobilized target DNAs arrayed in small spots. High-copy-number amplifications contained in tumor cells were rapidly scored by use of target DNAs as small as a cosmid. Low-copy-number gains and losses were identified reliably by their ratios by use of chromosome-specific DNA libraries or genomic fragments as small as 75 kb cloned in P1 or PAC vectors as targets, thus greatly improving the resolution achievable by chromosomal CGH. The ratios obtained for the same chromosomal imbalance by matrix CGH and by chromosomal CGH corresponded very well. The new matrix CGH protocol provides a basis for the development of automated diagnostic procedures with biochips designed to meet clinical needs. Genes Chromosomes Cancer 20:399–407, 1997. © 1997 Wiley-Liss, Inc.     

Genetic differences in endocrine pancreatic tumor subtypes detected by comparative genomic hybridization

The molecular pathogenesis as well as histogenesis of endocrine pancreatic tumors (EPTs) is not well understood, and the clinical behavior of EPTs is difficult to predict using current morphological criteria. Thus, more accurate markers of risk and better understanding of tumor initiation and progression are needed to allow a precise classification of EPTs. We have studied 44 benign and malignant EPTs by comparative genomic hybridization to correlate the overall number of genetic alterations with clinical and histopathological parameters and to identify chromosomal regions which might harbor genes involved in EPT pathogenesis and progression. Aberrations were found in 36 EPTs, and chromosomal losses (mean, 5.3) were slightly more frequent than gains (mean, 4. 6). The most frequent losses involved Y (45% of male EPTs), 6q (39%), 11q (36%), 3p, 3q, 11p (each 30%), 6p (27%), and 10q and Xq (each 25%), whereas most common gains included 7q (43%), 17q (41%), 5q and 14q (each 32%), 7p, 9q, 17p, 20q (each 27%), and 12q and Xp (each 25%). A correlation was found between the total number of genetic changes per tumor and both tumor size and disease stage. In particular, losses of 3p and 6 and gains of 14q and Xq were found to be associated with metastatic disease. Furthermore, characteristic patterns of genetic changes were found in the various EPT subtypes, eg, 6q loss in malignant insulinomas, indicating that these groups might evolve along genetically different pathways. The highlighted genetic aberrations, including the newly found involvement of 6q losses and sex chromosome alterations, should stimulate the further analysis of these chromosomal regions, which may lead to the discovery of novel genes important in the tumorigenesis and evolution of EPTs.     

Optimizing DOP-PCR for universal amplification of small DNA samples in comparative genomic hybridization

The standard comparative genomic hybridization (CGH) protocol relies on availability of macroscopic tumor samples, which do not contain too much interfering normal cells. Recently, CGH after universal amplification of genomic DNA with degenerate oligonucleotide primed PCR (DOP-PCR) has been used to detect genetic aberrations in microdissected tumor specimens. However, owing to the technical difficulties, CGH results of only few microdissected samples have so far been published. We have developed an improved protocol for DOP-PCR, which includes direct incorporation of fluorochrome-conjugated nucleotides into the PCR product. Among the four polymerase enzymes tested, ThermoSequenase gave the best yield, with PCR products ranging from 100–4,000 bp. A two-step PCR-procedure was used, consisting of a preamplification with low stringency conditions followed by amplification in more stringent conditions. The method was first validated by hybridizing DOP-PCR-amplified normal DNA against nick-translated reference DNA, which showed uniform and even hybridization result for all chromosomes. Comparison of DOP-PCR CGH to conventional CGH in MCF-7 breast cancer cell line further indicated that genetic aberrations can be reliably detected after DOP-PCR amplification. The sensitivity of the DOP-PCR-CGH was tested by serial dilution of MCF-7 DNA. Fifty picograms of sample DNA (corresponding roughly to two MCF-7 cells) was sufficient for high quality CGH. Experiments with cells microdissected from intraductal breast cancer demonstrated that carcinoma cells from 1 to 2 ducts were sufficient for a successful DOP-PCR CGH analysis. We conclude that the improved DOP-PCR-CGH protocol provides a powerful tool to study genetic aberrations in different histological subpopulations of malignant as well as precancerous lesions. DOP-PCR also improves the success rate of conventional paraffin-block CGH, because a poor quality or a too low yield of extracted DNA can be compensated by universal DNA amplification by DOP-PCR. Genes Chromosom. Cancer 18:94–101, 1997. © 1997 Wiley-Liss, Inc.     

Fluorescent in situ hybridization and array comparative genomic hybridization: complementary techniques for genomic evaluation

During the past few years a new high-throughput molecular technology, array comparative genomic hybridization, has received a great deal of attention. As a DNA-based tool, this technique is presumably more reproducible than expression arrays. In this review, I discuss how array comparative genomic hybridization is remarkably similar with regard to genome analysis to fluorescent in situ hybridization, a technique that is generally regarded as one of the more accurate and reproducible molecular techniques in diagnostic surgical pathology. A thorough understanding of this technology will be useful for all surgical pathologists in the near future, as this technology will no doubt have some influence on our daily practice.     

髓母细胞瘤染色体DNA失衡的比较基因组杂交研究

目的 探讨髓母细胞瘤染色体基因组DNA失衡及其与患者年龄、性别之间的关系.方法 用比较基因组杂交方法对16例髓母细胞瘤的染色体基因组DNA获得和丢失进行检测.结果 16例髓母细胞瘤中,共有15例(15/16)检测到获得和(或)丢失.有获得者10例(10/16),有丢失者11例(11/16),二者的差异无统计学意义(P＞0.05);获得和丢失例数的性别及年龄差异也无统计学意义(P＞0.05).出现单染色体、双染色体、三染色体及多染色体获得和(或)丢失者分别为3例(3/15),4例(4/15),1例(1/15)和7例(7/15).该组病例共检测到11个有DNA获得(+5q、+6q、+7q、+1lq、+15q、+17p、+17q、+19q、+20q、+2lq、+Xp)和25个有DNA丢失(-1p、-1q、-2p、-2q、-3q、-4p、-6p、-6q、-8p、-8q、-10p、-10q、-11p、-14q、-16p、-16q、-17p、-18p、-18q、-19p、-19q、-20p、-20q、-Xp、-Xq)的染色体区带;以+7q(6/16)、+17q(6/16)、-14q(5/16)和-10q(3/16)最常见;且-14q均发生在＞10岁组.结论 大多数髓母细胞瘤有不同程度的染色体基因组DNA失衡,常见失衡区带主要位于染色体长臂,+7q、+17q、-14q和-10q与该肿瘤的发生密切相关,-14q是导致＞10岁组髓母细胞瘤发生的重要因素,髓母细胞瘤可能存在不同的分子遗传学亚型.     

Aneuploidy detection in single cells using DNA array-based comparative genomic hybridization

The use of metaphase comparative genomic hybridization (CGH) to screen all human chromosomes for aneuploidy in preimplantation embryos is hindered by the time required to perform the analysis. We report in this paper a novel approach to manufacture a DNA microarray for CGH for the detection of aneuploidy in single cells. We spotted human chromosome-specific libraries on glass slides that were depleted of repetitive sequences and tested our array CGH method in 14 experiments using either single male and/or single female lymphocytes. For the autosomes, the mean normalized ratios were all close to the expected ratio of 1.0 with overall 300/308 (97%) of the normalized ratios falling within the range 0.75 to 1.25. It was possible to deduce the correct copy number of the X chromosome in 13/14 (92.9%) separate array CGH experiments but the Y chromosome in only 4/14 (29%). We tested our microarray CGH method on a single fibroblast from each of three cell lines containing a specific chromosome aneuploidy (trisomy 13, 15 or 18) and in each case our microarray analysis was able to obtain a diagnosis based on the fact that the aneuploid chromosome gave the highest ratio (1.32, 1.27 and 1.27 respectively) with the ratios of all other chromosomes falling within the range 0.75-1.25. Requiring just 30 h, our method may be more suitable for PGD aneuploidy screening than metaphase CGH.     

Universal linkage system: an improved method for labeling archival DNA for comparative genomic hybridization.

Comparative genomic hybridization (CGH) has become a powerful technique for studying gains and losses of DNA sequences in solid tumors. Importantly, DNA derived from archival tumor tissue is also applicable in CGH analysis. However, DNA isolated from routinely processed, formalin-fixed, paraffin-embedded tissue is often degraded, with the bulk of DNA showing fragment sizes of only 400-750 bp. Enzymatic labeling of archival DNA by standard nick translation (NT) decreases DNA size even further, until it becomes too small for CGH (<300 bp). This study presents application in CGH of a commercially available, non-enzymatic labeling method, called Universal Linkage System (ULS), that leaves the DNA fragment size intact. To compare the effect of chemical labeling of archival DNA by ULS vs. enzymatic by NT on the quality of CGH, DNA derived from 16 tumors was labeled by both ULS and NT. In those cases (n = 8), in which the bulk of DNA had a fragment size of 400-1,000 bp, CGH was successful with ULS-labeled probes, but not with NT-labeled probes. In the DNA samples (n = 6) with a fragment size > 1 kb, the intensity of CGH signals was comparable for both ULS- and NT-labeled probes, but CGH with ULS-labeled samples showed a high, speckled, background, which seriously hampered image analysis. In the remaining two cases, which had evenly distributed DNA fragment sizes (range 250-5,000 bp), CGH was successful with both labeling methods. Using DNA fragment size < 1 kb as a selection criterion for ULS labeling, we were able to obtain good quality CGH of a large panel (n = 77) of a variety of archival solid tumors. We conclude that ULS is an excellent labeling method for performing CGH on small-fragment-sized DNA.     

Gains and losses of DNA sequences in liposarcomas evaluated by comparative genomic hybridization

Comparative genomic hybridization (CGH) was used to detect and map the regions of gain, high-level amplification, and loss of DNA sequences in 14 liposarcomas. Thirteen tumors showed DNA sequence copy number changes of one or more genomic regions (mean, six aberrations/tumor; range, 0–17). These aberrations were observed in almost every chromosome but some chromosomal regions were affected more often than others. DNA sequence gains were more frequent than losses. The most common gain was seen at 12q14-21 (50% of tumors). Other frequent gains (29%) were of Iq21-24, 8cen-q21.2, 19q, and 20q. High-level amplification was observed in six (43%) tumors and included as minimal common segments bands 12q15, Iq22, and Iq24. In five (36%) tumors, sequences at Iq21-24 and Iq32 were found to be gained simultaneously with 12q14-21, which means that in 71% of the tumors with gain at 12q, an increase of DNA sequence copy number at Iq was also observed. The most common losses of DNA sequences (21%) occurred from regions 9p21-pter and 13q21-qter. Most of the aforementioned regions have not previously been reported to be altered in liposarcomas. The detection of a novel recurring amplicon at Iq21-24 with high-level amplification at Iq22 and frequent simultaneous DNA sequence gain at 12q14-21 (high-level amplification at 12q15) suggests that genes linked to both these regions may play a significant role in the development and progression of liposarcomas. Genes Chromosom Cancer 15:89–94 (1996). © 1996 Wiley-Liss, Inc.     

Molecular cytogenetic differences between histological subtypes of malignant mesotheliomas: DNA cytometry and comparative genomic hybridization of 90 cases

It is established that subtypes of human malignant mesotheliomas (MM) are associated with different survival times. Ninety cases of MM were examined using DNA cytometry and comparative genomic hybridization (CGH), with emphasis on the main histological subtypes; epithelioid, sarcomatoid and biphasic. A comparison by DNA cytometry revealed moderate differences, with the rare subgroup of mesodermomas having the highest and the sarcomatoid group the lowest rate of aneuploidy. Using CGH, 6.2 chromosomal imbalances per case on average could be detected. Losses (4.1/case) were more common than gains of chromosomal material (2.1/case). MM show no single, specific defect, but a typical pattern of genomic defects can be attributed to this tumour entity. Common losses are clustered at the chromosomal regions 9p21 (34%), 22q (32%), 4q31-32 (29%), 4p12-13 (25%), 14q12-24 (23%), 1p21 (21%), 13q13-14 (19%), 3p21, 6q22, 10p13-pter and 17p12-pter (16% each). Common gains are located on 8q22-23 (18%), 1q23/1q32 (16%), 7p14-15 and 15q22-25 (14% each). While differences in the frequencies of the defects between epithelioid and sarcomatoid MM are not as pronounced as are seen with the pleomorphic mesodermomas, several chromosomal locations (3p, 7q, 15q, 17p) show significant variations. The most pronounced distinguishing feature of sarcomatoid MM is a more than fourfold higher number of amplicons. These data indicate that MM has a distinctive tumour biology with a broad spectrum of heterogeneity, as reflected in morphology and also, more subtly, in the patterns of chromosomal imbalances of the subtypes.     

DNA copy number amplifications in human neoplasms: review of comparative genomic hybridization studies.

This review summarizes reports of recurrent DNA sequence copy number amplifications in human neoplasms detected by comparative genomic hybridization. Some of the chromosomal areas with recurrent DNA copy number amplifications (amplicons) of 1p22-p31, 1p32-p36, 1q, 2p13-p16, 2p23-p25, 2q31-q33, 3q, 5p, 6p12-pter, 7p12-p13, 7q11.2, 7q21-q22, 8p11-p12, 8q, 11q13-q14, 12p, 12q13-q21, 13q14, 13q22-qter, 14q13-q21, 15q24-qter, 17p11.2-p12, 17q12-q21, 17q22-qter, 18q, 19p13.2-pter, 19cen-q13.3, 20p11.2-p12, 20q, Xp11.2-p21, and Xp11-q13 and genes therein are presented in more detail. The paper with more than 150 references and two tables can be accessed from our web site http://www.helsinki.fi/lglvwww/CMG.html. The data will be updated biannually until the year 2001.     

Optimizing comparative genomic hybridization for analysis of DNA sequence copy number changes in solid tumors

Comparative genomic hybridization (CGH) is a powerful new method for molecular cytogenetic analysis of cancer. In a single hybridization, CGH provides an overview of DNA sequence copy number changes (losses, deletions, gains, amplifications) in a tumor specimen and maps these changes on normal chromosomes. CGH is based on the in situ hybridization of differentially labeled total genomic tumor DNA and normal reference DNA to normal human metaphase chromosomes. After hybridization and fluorescent staining of the bound DNAs, copy number variations among the different sequences in the tumor DNA are detected by measuring the tumor/normal fluorescence intensity ratio for each locus in the target metaphase chromosomes. CGH is in particular useful for analysis of DNA sequence copy number changes in common solid tumors where high-quality metaphase preparations are often difficult to make, and where complex karyotypes with numerous markers, double minutes, and homogeneously stained chromosomal regions are common. CGH only detects changes that are present in a substantial proportion of tumor cells (i.e., clonal aberrations). It does not reveal translocations, inversions, and other aberrations that do not change copy number. At present, CGH is a research tool that complements previous methods for genetic analysis. CGH will advance our understanding of the genetic progression of cancer and highlight important genomic regions for further study. Direct clinical applications of CGH are possible, but will require further development and validation of the technique. We describe here our recent optimized procedures for CGH, including DNA labeling, hybridization, fluorescence microscopy, digital image analysis, data interpretation, and quality control, emphasizing those steps that are most critical. We will also assess sensitivity and resolution limits of CGH as well as discuss possible future technical improvements. Genes Chromosom Cancer 10:231–243 (1994). © 1994 Wiley-Liss, Inc.     

DNA microarrays for comparative genomic hybridization based on DOP-PCR amplification of BAC and PAC clones

We have designed DOP-PCR primers specifically for the amplification of large insert clones for use in the construction of DNA microarrays. A bioinformatic approach was used to construct primers that were efficient in the general amplification of human DNA but were poor at amplifying E. coli DNA, a common contaminant of DNA preparations from large insert clones. We chose the three most selective primers for use in printing DNA microarrays. DNA combined from the amplification of large insert clones by use of these three primers and spotted onto glass slides showed more than a sixfold increase in the human to E. coli hybridization ratio when compared to the standard DOP-PCR primer, 6MW. The microarrays reproducibly delineated previously characterized gains and deletions in a cancer cell line and identified a small gain not detected by use of conventional CGH. We also describe a method for the bulk testing of the hybridization characteristics of chromosome-specific clones spotted on microarrays by use of DNA amplified from flow-sorted chromosomes. Finally, we describe a set of clones selected from the publicly available Golden Path of the human genome at 1-Mb intervals and a view in the Ensembl genome browser from which data required for the use of these clones in array CGH and other experiments can be downloaded across the Internet.     

Detection of DNA copy number alterations in cancer by array comparative genomic hybridization

Over the past few years, various reliable platforms for high-resolution detection of DNA copy number changes have become widely available. Together with optimized protocols for labeling and hybridization and algorithms for data analysis and representation, this has lead to a rapid increase in the application of this technology in the study of copy number variation in the human genome in normal cells and copy number imbalances in genetic diseases, including cancer. In this review, we briefly discuss specific technical issues relevant for array comparative genomic hybridization analysis in cancer tissues. We specifically focus on recent successes of array comparative genomic hybridization technology in the progress of our understanding of oncogenesis in a variety of cancer types. A third section highlights the potential of sensitive genome-wide detection of patterns of DNA imbalances or molecular portraits for class discovery and therapeutic stratification.     

Detection of DNA gains and losses in primary endometrial carcinomas by comparative genomic hybridization

Comparative genomic hybridization (CGH) was used in a retrospective analysis of chromosomal imbalances in frozen primary tumor specimens from 14 endometrial carcinoma patients. Chromosome changes were detected in nine cases (64%), and tumor stage and grade tended to parallel the degree of genomic imbalances. Gain of the entire long arm of chromosome I was observed in six cases (43%), three of which displayed only this chromosome change. Other common sites of copy number increases included 8q21 → qter (4 cases), 10p15 (4 cases), 10q11 → q24 (3 cases), and 13q21 → qter (3 cases, each with stage III disease). Two of the tumors with gains of chromosome 10 involved the whole chromosome, and this was the sole abnormality in one case. DNA amplification at 5p14 → p15 was identified in one specimen, a stage III tumor having numerous imbalances. DNA microsatellite analysis revealed multiple replication errors (RER), indicative of the RER⁺ phenotype, in four of 13 (31%) cases evaluated. The RER⁺ phenotype was observed in four of six stage Ia tumors but in none of seven stage Ib or stage III tumors. Multiple genomic imbalances detected by CGH were not observed in RER⁺ tumors but were detected in five of nine tumors without the RER⁺ phenotype. These investigations demonstrate the feasibility of CGH for the retrospective assessment of chromosomal changes in endometrial carcinoma specimens. Moreover, these data suggest that the etiologies in tumors with and without the RER⁺ phenotype may differ. Genes Chromosom. Cancer 18:115–125, 1997. © 1997 Wiley-Liss, Inc.     

DNA copy number changes in epithelioid sarcoma and its variants: a comparative genomic hybridization study.

Epithelioid sarcoma is a distinctive, rare soft tissue sarcoma that typically involves the distal extremities in young adults, and shows epithelioid morphology and immunohistochemical markers of epithelial differentiation. The genetic background of epithelioid sarcoma is poorly understood, and knowledge of it could give insights into the pathogenesis of this tumor and its possible relationship with other malignant tumors. In this study, we analyzed DNA copy number changes in 30 epithelioid sarcomas by comparative genomic hybridization. DNA was extracted from microdissected samples of formaldehyde-fixed and paraffin-embedded tumors with a minimum of 60% of tumor cells in each sample. Sixteen tumors (53%) showed DNA copy number changes at one to six different genomic sites. The majority of the changes were gains, seen in 14 tumors, whereas 10 tumors showed losses. The most common recurrent gains were at 11q13 (five cases), 1q21-q23 (four cases), 6p21.3 (three cases), and 9q31-qter (three cases). High-level amplifications were detected once in 6p21.3-p21.1 and once in 9q32-qter. Recurrent losses were seen at 9pter-p23 (three cases), 13q22-q32 (three cases), 1p13-p22 (two cases), 3p12-p14 (two cases), 4q13-q33 (two cases), 9p21 (two cases), and 13q32-qter (two cases). The most common recurrent gain at 11q13 was seen in both classic cases and angiomatoid and rhabdoid variants supporting the relationship of these variants with the classic epithelioid sarcoma. Expression of cyclin D1 gene, located in 11q13, was immunohistochemically detected in nine of 15 cases including three of five cases with gain of 11q13, suggesting its involvement in epithelioid sarcoma. The observed comparative genomic hybridization changes give targets for future genetic studies on epithelioid sarcoma.     

Gains, losses, and amplifications of DNA sequences evaluated by comparative genomic hybridization in chondrosarcomas.

Comparative genomic hybridization was used to search for previously unknown gains and losses of DNA sequences along all chromosome arms in 29 chondrosarcoma specimens obtained from 23 patients. Extensive genetic aberrations, with a mean of 6 changes per tumor (range, 1 to 24), were detected in 21 of the 29 samples analyzed (72%). The majority of these changes were gains of whole chromosomes or whole chromosome arms. Gains of DNA sequence copy number were most frequent at 20q (38%), 17p (38%), 20p (31%), 1cen-q24 (28%), and 14q23-qter (28%). High-level amplifications of small chromosome regions were sporadic, detected in only 17% of the samples. The only recurrent high-level amplification, seen in two tumors (7%), affected the minimal common region 12cen-q15. Other amplifications, each encountered only once, involved 1p33-p35, 2p23-pter, 4p, 6p22-pter, 18q12-q22, 19p13.2, 19q13.2, and 20q13.1. Losses of DNA sequences were rare and were most commonly observed at 6cen-q22 (17%) and 9p (17%).     

Recurrent DNA copy number changes revealed by comparative genomic hybridization in primary Merkel cell carcinomas.

Comparative genomic hybridization (CGH) was used to search for gains, high-level amplifications and losses of DNA sequences along all chromosome arms in 19 primary Merkel cell carcinomas (MCC). Extensive genetic aberrations, with a mean value of 5.5+/-1.1 changes per tumor were detected in 13 out of the 19 samples analyzed. Our CGH results reveal several new and other previously known chromosomal regions that are involved in the pathogenesis of MCC. The majority of the alterations were gains of whole chromosomes or whole chromosome arms. Compared to losses, the frequency of DNA copy number gains was two-fold. DNA sequence copy number gains were most common in chromosomes 6 (42%), 1 (37%), and 5 (32%). The most frequent minimal common regions of gains were 6pterqter (42%), 1q11q31 (32%), and 5p (32%). No recurrent high-level amplifications were observed. High-level amplifications of small chromosomal regions were found in four samples out of the 19 tumors analyzed (21%). Amplifications affected 1q22q24 (5%), 4p (5%), and 5p (5%). Losses most frequently affected chromosomes 13 (21%) and 4 (16%). Minimal common regions with the most frequent losses were 13q13q31 (21%), 4q (16%), and 16q (11%). No significant statistical correlation between genomic aberrations and clinicopathological factors was revealed, despite the fact that there was an obvious tendency towards it. Primary MCC expressing DNA alterations were predominantly distinguished in large tumors, and risk of metastatic dissemination was three-fold compared to tumors with no DNA alterations.     

Analysis of ovarian borderline tumors using comparative genomic hybridization and fluorescence in situ hybridization.

It is unclear whether ovarian borderline tumors (tumors of low malignant potential) are independent entities or whether they are part of a continuum of tumor progression that culminates in ovarian carcinoma. Little is known about genetic abnormalities in borderline tumors because of the difficulty of growing them in culture for chromosome studies, and because the low ratio of tumor to nontumor cells can interfere with molecular genetic examination. To circumvent these problems, we performed comparative genomic hybridization (CGH) on 10 serous borderline tumors from nine patients, using microdissection to enrich the samples for tumor DNA and reduce contamination from stromal and inflammatory cells. CGH analysis revealed that three of the tumors had detectable chromosomal imbalances, whereas seven were in a balanced state. In those tumors with imbalances, the number of abnormalities ranged from 3-6 per tumor. Additional studies by fluorescence in situ hybridization (FISH) on disaggregated nuclei confirmed the imbalances detected by CGH, revealed one tumor to be hypertriploid, and indicated that the remaining tumors were diploid and in a balanced state. All abnormalities observed in the aneuploid cases are consistent with chromosomal aberrations previously reported for ovarian carinomas, providing further evidence that some borderline tumors are part of a continuum of tumor progression. These results also suggest that there may be different mechanisms leading to borderline tumor formation, including one associated with multiple chromosomal imbalances, and others that do not involve imbalances detectable by CGH. Genes Chromosomes Cancer 25:307-315, 1999.