Comparative cytogenetic and DNA flow cytometric analysis of 242 primary breast carcinomas

The cytogenetic and DNA flow cytometric findings in 242 breast carcinomas were compared. The combined use of both techniques improved the detection of abnormal cell populations from 65% by cytogenetic analysis alone and 59% by DNA flow cytometric analysis alone to 84%. Informative and comparable cytogenetic and flow cytometric data were obtained for 155 tumors. Among these 155 tumors, there was good concordance (64%) between the estimates of genomic changes by the two methods. Most discrepancies were among the DNA-diploid cases, where cytogenetic analysis detected small genomic changes. There were, however, also some exceptions in which large genomic changes detected by one method were missed by the other. Of the specific breast cancer-associated cytogenetic aberrations subjected to separate correlation analysis, polysomy for chromosome 20 was significantly associated with a high S-phase fraction, whereas loss of the long arm of chromosome 16 and/or the presence of a der(1;16) were significantly associated with a low S-phase fraction. Our data show that cytogenetic and DNA flow cytometric analyses of breast carcinomas give largely comparable results, and that combining data from both methods significantly improves the information obtained by either technique used alone on the genetic abnormalities in these tumors.

CRD-BP: a c-Myc mRNA stabilizing protein with an oncofetal pattern of expression

The Coding Region Determinant-Binding Protein (CRD-BP) is an RRM and KH-domain-containing protein that recognizes specifically at least three RNAs. It binds to one of the two c-myc mRNA instability elements, to the 5'Un Translated Region (UTR) of the leader 3 IGF-II mRNA and to the oncofetal H19 RNA. CRD-BP has been assigned a role in stabilizing c-myc mRNA by preventing its endonucleolytic cleavage and in repressing the translation of the leader 3 IGF-II mRNA, the major embryonic species of this message. CRD-BP is normally expressed only in fetal tissues. However, its expression is detected in primary tumors and transformed cell lines of different origins. The vast majority of colon (80%) and breast (60%) tumors and sarcomas (73%) express CRD-BP whereas in other tumor types, for example prostate carcinomas, its expression is rare. CRD-BP expression has also been detected in benign tumors such as breast fibroadenomas, meningiomas and other benign mesenchymal tumors, implying a role for this gene in abnormal cell proliferation. In breast carcinomas, CRD-BP expression and or gene copy number gains in the region encompassing the c-myc locus were detected in approximately 75% of tumors, implying that the deregulated expression of c-myc may be more widespread than previously believed. Infiltrated lymph nodes, corresponding to CRD-BP-positive primary tumors, were also found positive indicating that monitoring for CRD-BP could prove useful for the detection and monitoring of disseminated disease.

8q24 Copy number gains and expression of the c-myc mRNA stabilizing protein CRD-BP in primary breast carcinomas

The coding region determinant binding protein (CRD-BP) was isolated by virtue of its high affinity to the c-myc mRNA coding region stability determinant and shown to shield this message from nucleolytic attack, prolonging its half-life. CRD-BP is normally expressed during fetal life but is also activated de novo in tumors. Considering that aberrant CRD-BP expression may represent an additional mechanism interfering with c-myc regulation, we screened 118 primary breast carcinomas for CRD-BP expression, 60 of which had also been analyzed by comparative genomic hybridization (CGH). Copy number gains encompassing 8q24, the chromosome band that contains the c-myc locus, were detected in 48.3% (29/60) of tumors, whereas gains involving band 17q21, which contains the CRD-BP locus, were observed in 18.3% (11/60) of tumors. CRD-BP expression was detected in 58.5% (69/118) of tumors, implying mechanisms of activation alternative to gene amplification. Altogether, some 75% of the tumors had alterations pertaining to c-myc since they either harbored 8q24 gains and/or expressed CRD-BP. Significant associations were detected between CRD-BP expression and the absence of estrogen receptors (p = 0.005) and between the presence of 8q24 gains and an increased number of genomic changes as measured by CGH (p = 0.0017). Tumors were divided into 4 groups according to CRD-BP expression and 8q24 gains. The odds for tumors having both characteristics to be classified as poorly differentiated (grade III vs. grade I and II) were 19.6 times the corresponding odds for tumors neither expressing CRD-BP nor harboring 8q24 gains. For tumors either harboring 8q24 gains only or expressing CRD-BP alone, the corresponding odds were 6.4 and 3, respectively.

Telomerase activity and genetic alterations in primary breast carcinomas

It has been proposed that the structural and numerical chromosome abnormalities recorded in breast cancer could be the result of telomere dysfunction and that telomerase is activated de novo to provide a survival mechanism curtailing further chromosomal aberrations. However, recent in vivo and in vitro data show that the ectopic expression of telomerase promotes tumorigenesis via a telomere length-independent mechanism. In this study, the relation between telomerase expression and the extent of chromosomal aberrations was investigated in 62 primary breast carcinomas. Telomerase activity was measured using a polymerase chain reaction-based telomeric repeat amplification protocol assay and 92% of the tumors were found to express telomerase with a relative activity ranging from 0 to 3839.6. Genetic alterations were determined by G-banding and comparative genomic hybridization analysis and 97% of the tumors exhibited chromosomal aberrations ranging from 0 to 44 (average: 10.98). In the overall series, the relationship between telomerase activity levels and genetic changes could be best described by a quadratic model, whereas in tumors with below-average genetic alteration numbers, a significant positive association was recorded between the two variables (coefficient=0.374, P=.017). The relationship between telomerase activity levels and the extent of genetic alteration may reflect the complex effect of telomerase activation upon tumor progression in breast carcinomas.

Cytogenetic profile of unknown primary tumors: clues for their pathogenesis and clinical management

Unknown primary tumors (UPTs) represent an entity of great clinical and biological interest, whose origin cannot be determined even after medical workup. To better understand their pathogenesis by outlining their genetic composition, 20 UPTs were investigated by G-banding, supplemented with Fluorescence In Situ Hybridization and Comparative Genomic Hybridization analyses. The data obtained were sufficient to reach a diagnosis in five cases-four lymphomas and one Ewing sarcoma-demonstrating that in a subset of UPTs, cytogenetics can be an adjunct for differential diagnosis. In the remaining 15 UPTs, an aggressive cytogenetic pattern was revealed. The most frequently rearranged chromosome regions were 1q21, 3p13, 6q15-23, 7q22, 11p12-5, and 11q14-24, pinpointing gene loci probably associated with the peculiar pathogenesis of UPTs. The preferential involvement of 4q31, 6q15, 10q25, and 13q22 in adenocarcinomas (whereas 11q22 is involved in the rest of the carcinomas)-in addition to the marked divergence in the mean average of chromosomal changes, 16 and 3, respectively-demonstrates genotypic differences between the two histologic subgroups. Furthermore, the significantly shorter survival in cases displaying massive chromosome changes compared with those having a few changes indicates that the cytogenetic pattern might be used as a tool to assess prognosis in UPTs, even without the detection of their primary site.

Germ line BRCA1 & BRCA2 mutations in Greek breast/ovarian cancer families: 5382insC is the most frequent mutation observed

BRCA1 and BRCA2 genes were screened for loss-of-function mutations in a series of 85 patients having at least one first- or second-degree relative affected by breast and/or ovarian cancer. All BRCA1 exons and BRCA2 exons 10 and 11 were screened with a combination of methods including SSCP, PTT and direct sequencing. We have found disease-associated mutations in 14 families (16.5%), eleven in BRCA1 and three in BRCA2. The known founder mutation 5382insC of BRCA1 was identified in seven unrelated families. The other mutations identified include the non-sense R1751X, the splice junction variant 5586G>A of BRCA1 and three frameshifts, 2024del5, 3034del4, and 6631del5, of BRCA2. Nine out of these 14 families had a family history of three or more breast/ovarian cancer cases. A large number of polymorphic or unclassified variants is also reported. Combined with our previously published data 5382insC was found in nine out of 20 families (45%), suggesting that this mutation may represent a common founder mutation in the Greek population.

Genome profiling of breast cancer cells selected against in vitro shows copy number changes

About 20% of breast carcinomas show no clonal chromosome abnormalities when analyzed after short-term culturing. An interesting question is whether this subset of breast carcinomas really is karyotypically normal or if selection for normal cells occurred in vitro. To address this issue, 26 breast carcinomas that had shown no cytogenetic changes by chromosome banding analysis were examined by comparative genomic hybridization (CGH), a technique that does not require culturing or tumor metaphase cells. All but one case showed copy number changes by CGH (median, four). A comparison of these findings with those of a karyotypically abnormal series analyzed using the same CGH protocol found that the cytogenetically "normal" cases were typically genetically less complex (median, four and eight, respectively; P = 0.0058). Although largely the same alterations were found in both series, some differences with respect to the frequencies of specific imbalances were seen. Gains of 3p and 6q and losses of 10q, 14q, and 17p more often were found in the cytogenetically abnormal series than in the normal tumors. We conclude that in most instances cells found to be normal by chromosome banding analysis after short-term culture do not belong to the tumor parenchyma. Furthermore, when we compared the distribution of the number of imbalances detected by CGH in the total data set according to the mitotic index in vivo (scored from 1 to 3), the median values were three, seven, and 18, respectively (P < 0.001). These data indicate not only that karyotypically normal breast carcinomas may represent a genetically simpler subgroup that grows poorly in vitro but also that this subset of tumors already has a slow growth rate in vivo.

Differentially amplified chromosome 12 sequences in low- and high-grade osteosarcoma

Most osteosarcomas are highly aggressive malignancies characterized by a complex pattern of chromosome abnormalities. However, a subgroup of low-grade, parosteal tumors exhibits a relatively simple aberration pattern dominated by ring chromosomes carrying amplified material from chromosome 12. To assess whether sequences from this chromosome were differentially amplified in low- and high-grade osteosarcomas, copy numbers of the CCND2, ETV6, KRAS2, and D12S85 regions in 12p and the MDM2 region in 12q were evaluated by interphase or metaphase fluorescence in situ hybridization (FISH) in 24 osteosarcomas. Amplification of MDM2 was detected in all five low-grade and four high-grade osteosarcomas, all of which showed ring chromosomes. An overrepresentation of 12p sequences was found in 1/5 low-grade and in 9/19 high-grade tumors. Multicolor single-copy FISH analysis of metaphase cells from six high-grade tumors showed that extra 12p material either occurred together with MDM2 in ring chromosomes or was scattered over the genome as a result of complex structural rearrangements. Most tumors (8/10) not containing amplification of the assessed chromosome 12 loci exhibited a nondiploid pattern at evaluation with probes for centromeric alpha satellite sequences. These findings indicate that gain of sequences from the short arm of chromosome 12 could be a possible genetic pathway in the development of aggressive osteosarcoma.

Cytogenetic clues to breast carcinogenesis

The somatic mutation theory of cancer maintains that tumorigenesis is driven by genetic alterations, many of which are visible cytogenetically. We have examined breast cancer by chromosome banding analysis after short-term culturing of tumor cells and here review our findings in 322 karyotypically abnormal samples obtained since 1992 from 256 patients. The screening capabilities of this technique enabled us to identify several cytogenetic subgroups of breast cancer, to study the intratumor heterogeneity of breast carcinomas, and to compare primary tumors with their metastases. Using chromosome abnormalities as clonality markers, we could determine on an individual basis when multiple, ipsilateral or bilateral breast, tumors were independent de novo carcinomas and when they resulted from the spreading of a single malignant clone within one breast or from one breast to the other. The distribution of chromosomal breakpoints and genomic gains and losses is clearly nonrandom in breast cancer, something that can guide further investigations using molecular methods. Based on the total dataset, we propose a multipathway model of mammary carcinogenesis that takes into consideration the genetic heterogeneity revealed by the karyotypic findings and review the karyotypic-pathologic correlations and the possible clinical applications of the cytogenetic knowledge.

C-MYC and IGF-II mRNA-binding protein (CRD-BP/IMP-1) in benign and malignant mesenchymal tumors

Mouse coding region determinant-binding (mCRD-BP) and human IGF-II mRNA-binding 1 (hIMP-1) proteins are orthologous mRNA-binding proteins that recognize c-myc and IGF-II mRNA, respectively, and regulate their expression posttranscriptionally. Here, we confirm that human CRD-BP/IMP-1 binds to c-myc mRNA and that it is predominantly expressed in fetal tissues. Moreover, hCRD-BP/IMP-1 expression was detected in cell lines of neoplastic origin and in selected primary tumors. In a series of 33 malignant and 10 benign mesenchymal tumors, 73% and 40%, respectively, were found to express hCRD-BP/IMP-1. In particular, expression was significant in 14 Ewing's sarcomas, all of which were positive. The data suggest that hCRD-BP/IMP-1 plays a role in abnormal cell proliferation in mesenchymal tumors.

Are some breast carcinomas polyclonal in origin?

This editorial comments on the important study by Going et al. published in the present issue of the Journal [1]. Using a molecular genetic assay based on the X-chromosome inactivation principle, they found that 4 out of 12 breast carcinomas examined exhibited what the authors call "clonal mosaicism" that is, two or more monoclonal samples were mosaic (polyclonal) in respect of X chromosome inactivation between separate, morphologically homogeneous tumour areas. The authors very carefully discuss potential methodological errors that could have led to this surprising finding, which seems to run counter to the almost unanimously held conviction that carcinomas are monoclonal in origin, but none of these potential errors would explain the results. As often in such situations, the authors prudently state that further studies using independent analytical techniques are necessary to find out whether a significant proportion of mammary carcinomas are indeed polyclonal. However, there already exists a substantial body of evidence from cytogenetic studies of breast cancers indicating that many of them are polyclonal. Although there is still room for interpretation and some doubt remains as to exactly which role should be ascribed to the observed clonal heterogeneity in our models of carcinogenesis, it seems obvious that more attention than before ought to be paid to this now well documented fact.

Evaluation of breast cancer polyclonality by combined chromosome banding and comparative genomic hybridization analysis

Cytogenetically unrelated clones have been detected by chromosome banding analysis in many breast carcinomas. Because these karyotypic studies were performed on short-term cultured samples, it may be argued that in vitro selection occurred or that small clones may have arisen during culturing. To address this issue, we analyzed 37 breast carcinomas by G-banding and comparative genomic hybridization (CGH), a fluorescent in situ hybridization--based screening technique that does not require culturing or tumor metaphases. All but two of the 37 karyotypically abnormal cases presented copy number changes by CGH. The picture of genomic alterations revealed by the two techniques overlapped only partly. Sometimes the CGH analysis revealed genomic imbalances that belonged to cell populations not picked up by the cytogenetic analysis and in other cases, especially when the karyotypes had many markers and chromosomes with additional material of unknown origin, CGH gave a more reliable overall picture of the copy number gains and losses. However, besides sometimes revealing cell populations with balanced chromosome aberrations or unbalanced changes that nevertheless remained undetected by CGH, G-banding analysis was essential to understand how the genomic imbalances arose in the many cases in which both techniques detected the same clonal abnormalities. Furthermore, because CGH pictures only imbalances present in a significant proportion of the test sample, the very detection by this technique of imbalances belonging to apparently small, cytogenetically unrelated clones of cells proves that these clones must have been present in vivo. This constitutes compelling evidence that the cytogenetic polyclonality observed after short-term culturing of breast carcinomas is not an artifact.

High frequency of clonal chromosome abnormalities in prostatic neoplasms sampled by prostatectomy or ultrasound-guided needle biopsy

Cancer of the prostate remains poorly characterized cytogenetically. This is due in part to methodological problems and in part to the paucity of radical prostatectomies, until now the main source of material for cytogenetic analyses. We have improved existing techniques for the culturing of prostatic neoplasms removed by radical prostatectomy or sampled by ultrasound-guided needle biopsy. Successful short-term cultures were obtained from all 10 prostatectomy samples and from all 10 ultrasound-guided needle biopsies, always with a pure epithelial morphology. Of the 19 cases yielding a sufficient number of high-quality metaphases for chromosome banding analysis, the single atypical epithelial hyperplasia had a normal karyotype, whereas both prostatic intraepithelial neoplasias and 12 of 16 (75%) invasive carcinomas were shown to have clonal abnormalities. Ten of the 12 (83%) karyotypically abnormal invasive carcinomas presented structural chromosomal rearrangements. A recurrent deletion, del(10)(p13), was seen in three tumors; in one of them the terminal nature of the deletion was confirmed by two-color FISH. A del(17)(p11) was seen in one PIN lesion, but since the analysis of exons 4-8 of the TP53 tumor suppressor gene revealed no mutations, there probably was no inactivation of the second TP53 allele. Our study thus leads to the following main conclusions. First, better culturing methods allow the detection of abnormal karyotypes in a much higher percentage of prostatic neoplasms than has hitherto been possible. Second, ultrasound-guided needle biopsies of prostatic neoplasms are a sufficient source of material for cytogenetic analysis. Third, a terminal deletion of the short arm of chromosome 10, del(10)(p13), seems to identify a subgroup of prostatic cancer.

Cytogenetic heterogeneity and clonal evolution in synchronous bilateral breast carcinomas and their lymph node metastases from a male patient without any detectable BRCA2 germline mutation

Two synchronous bilateral breast carcinomas and their matched lymph node metastases from a 70-year-old man were cytogenetically analyzed. All four tumors were near-diploid, and except for the primary tumor from the right breast, had a 45,X,-Y clone in common. The loss of the Y chromosome was, however, common to all four tumors, whereas metaphase cells from peripheral blood lymphocytes showed a normal 46, XY chromosome complement. The primary tumor from the right breast was monoclonal, with loss of the Y chromosome and gain of 1q, whereas its metastasis had two related clones: the 45,X,-Y clone, and the other a more complex version of the clone in the primary tumor, with inv(3), -14, and del(16)(q13) as additional changes. The primary tumor from the left breast was polyclonal with three unrelated clones: 45,X,-Y/45,XY,-18/47,XY,+20, two of which were present in its metastasis. DNA flow cytometric studies showed diploidy for both primary tumors. No mutation in the BRCA2 gene was found on analysis of DNA from peripheral blood lymphocytes. The present findings show that del(16)(q13) is a recurrent finding among male breast carcinomas and that some of the primary cytogenetic abnormalities, as well as the pattern of chromosomal changes during the progression of sporadic breast carcinoma in the male, are similar to those in the female. In addition, the loss of the Y chromosome in the tumors but not in peripheral blood lymphocytes, suggests a possible role for this abnormality in the pathogenesis of male breast carcinoma.

Spectral karyotyping and chromosome banding studies of primary breast carcinomas and their lymph node metastases

Three primary breast tumors and their lymph node metastases were characterized by G-banding, spectral karyotyping (SKY), and fluorescence in situ hybridization (FISH). In each case, the karyotypic abnormalities detected were similar in the primary tumor and its matched metastasis. Two of the pairs had near-diploid karyotypes with three to four chromosomal aberrations, whereas the third pair had a near-pentaploid chromosome content and many marker chromosomes in the primary tumor and a near-tetraploid chromosome number with almost the same marker chromosomes in the metastasis. SKY and FISH confirmed the karyotypic similarities between the primary tumors and their metastases and, in addition, improved the identification and characterization of marker chromosomes. One of the tumor pairs with near-diploid karyotypes had gain of 8q, 16q, and 17q, whereas the other had gain of 1q and chromosome 8 material in the form of ring chromosomes. The third pair had more complex chromosomal translocations and numerical changes resulting in net gain of material from chromosomes X, 1, 2, 6, 7, 14, 16, 19, and 20, and chromosome arms 8q and 11q, as well as net loss of material from chromosomes 3, 13, 18, 21, and 22. The present study underscores the need to combine conventional chromosome banding and molecular cytogenetic techniques in the cytogenetic analysis of solid tumors.

Recurrent t(16;17)(q22;p13) in aneurysmal bone cysts

Aneurysmal bone cyst (ABC) is a benign bone lesion for which no previous cytogenetic data exist. We describe the finding of clonal chromosome aberrations in three tumors; two had a t(16;17)(q22;p13) as the sole anomaly, and the third had a del(16)(q22) as the only change. These findings show that somatic mutations contribute to the development of ABC and furthermore indicate that bands 16q22 and 17p13 may harbor genes of importance in this process.

Karyotypic evolution in breast carcinomas with i(1)(q10) and der(1;16)(q10;p10) as the primary chromosome abnormality

The pattern of clonal karyotypic evolution in breast carcinomas carrying an i(1q) or a der(1;16)(q10;p10) as the primary chromosome abnormality was assessed in a series of 42 tumors, including 8 described here for the first time, with either or both (3 tumors) of them defining cytogenetic features. Evidence of clonal evolution was seen in somewhat more than half of all cases in both subgroups. The secondarily acquired aberrations appeared to be nonrandom in distribution. This was especially so for structural rearrangements of 11q leading to loss of material from this arm, which were clearly more common in both subgroups than in karyotypically abnormal breast carcinomas in general. Other deviations from random were less certain but seemed to include the frequent occurrence of +20 in tumors with i(1q) and +7 in tumors with der(1;16)(q10;p10). That differences were observed between i(1q) carcinomas and der(1;16)(q10;p10) carcinomas with regard to their patterns of clonal evolution hints that the pathogenetic effect of the primary change in these two situations may be more than the mere gain of an extra copy of 1q.

Different patterns of chromosomal imbalances in metastasising and non-metastasising primary breast carcinomas

In an attempt to identify chromosomal abnormalities that may be associated with a metastatic phenotype, we investigated the pattern of chromosomal gains and losses in 66 node-positive and 63 node-negative primary breast carcinomas. For both subgroups of tumours, losses were more common than gains and the losses were most often the result of structural aberrations. The exceptions were the long arm of chromosome 1, and chromosomes 7, 8, 12, 18 and 20, which were more often gained than lost. Node-negative tumours were preferentially characterised by loss of 6q10-21 and loss of 16q, whereas loss of chromosome 18 was significant for node-positive tumours. Other aberrations that tended to be associated with one of the phenotypes, though not statistically significant, were gain of chromosome 18 and loss of chromosome 10 in node-negative tumours, and gain of chromosome 14 and loss of 12p in node-positive tumours. Our data show that there are differences among the genetic lesions present in node-negative and node-positive breast tumours. Int. J. Cancer (Pred. Oncol.) 84:370-375, 1999.

Different patterns of chromosomal imbalances in metastasising and non-metastasising primary breast carcinomas

In an attempt to identify chromosomal abnormalities that may be associated with a metastatic phenotype, we investigated the pattern of chromosomal gains and losses in 66 node-positive and 63 node-negative primary breast carcinomas. For both subgroups of tumours, losses were more common than gains and the losses were most often the result of structural aberrations. The exceptions were the long arm of chromosome 1, and chromosomes 7, 8, 12, 18 and 20, which were more often gained than lost. Node-negative tumours were preferentially characterised by loss of 6q10-21 and loss of 16q, whereas loss of chromosome 18 was significant for node-positive tumours. Other aberrations that tended to be associated with one of the phenotypes, though not statistically significant, were gain of chromosome 18 and loss of chromosome 10 in node-negative tumours, and gain of chromosome 14 and loss of 12p in node-positive tumours. Our data show that there are differences among the genetic lesions present in node-negative and node-positive breast tumours. Int. J. Cancer (Pred. Oncol.) 84:370-375, 1999.

Multiple polysomies in breast carcinomas: preferential gain of chromosomes 1, 5, 6, 7, 12, 16, 17, 18, and 19

Chromosome G-banding analysis of metaphase cells from 16 primary breast carcinomas revealed the presence of multiple polysomies in near-diploid as well as in polyploid cells. Chromosome 17 was preferentially gained in 7 tumors, followed in frequency by chromosomes 1, 12, and 19 (5 tumors each), and chromosomes 5, 6, 7, 16, and 18 (4 tumors each). Eleven of the 16 carcinomas had, apart from the clones exhibiting the numerical gains, other unrelated clones. Nine of these 11 cases had clones with structural chromosome aberrations, 5 of which had structural aberrations involving the short arm of chromosome 3. The biologic significance, if any, of this seemingly nonrandom coexistence of multiple polysomies with structural aberrations of 3p is at present not known. The pattern of numerical chromosome aberrations observed in the present study is comparable to previous results from fluorescence in situ hybridization (FISH) studies, with the use of centromeric probes on interphase cells. However, unlike FISH studies, which have been focused on chromosomes 1, 3, 7, 8, 11, 16, and 17, the cytogenetic results reveal that other chromosomes also may be nonrandomly gained as part of multiple polysomies in breast carcinomas. In addition, the tumors with multiple polysomies were generally of high histologic grade and with metastasis to axillary lymph nodes, suggesting that multiple wholechromosome gains may be a pathway of genetic evolution or progression or both in some breast carcinomas.

Chromosomal aberrations in breast cancer: a comparison between cytogenetics and comparative genomic hybridization

The analysis of chromosomal imbalances in solid tumors using comparative genetic hybridization (CGH) has gained much attention. A survey of the literature suggests that CGH is more sensitive in detecting copy number aberrations than is karyotyping, although careful comparisons between CGH and cytogenetics have not been performed. Here, we compared cytogenetics and CGH in 29 invasive breast cancers after converting the karyotypes into net copy number gains and losses. We found 15 tumors (56%) with a significant agreement between the two methods and 12 tumors (44%) where the methods were in disagreement (two cases failed CGH analysis). Interestingly, in 13 of the 15 tumors where the two methods were concordant, there was also a strong correlation between chromosome index and DNA index by flow cytometry. In the opposite situation, i.e., when chromosome and DNA indices were not matching, there was disagreement between cytogenetics and CGH in 10 of the 12 tumors. Of the discordant cases, all except one had a "simple" abnormal karyotype. Unresolved chromosomal aberrations (marker chromosomes, homogeneously staining regions, double minutes) could not completely explain the differences between CGH and karyotyping. A likely explanation for the discrepancies is that the methods analyzed different cell populations. Gains and losses found by CGH represented the predominant (often aneuploid) clone, whereas the abnormal, near-diploid karyotypes represented minor cell clone(s), which, for unknown reasons, had a growth advantage in vitro.