Cancer Research Cancer Epigenetics EMT and Cancer Progression and Treatment
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nessling, M.
Right arrow Articles by Lichter, P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nessling, M.
Right arrow Articles by Lichter, P.
[Cancer Research 65, 439-447, January 15, 2005]
© 2005 American Association for Cancer Research

Molecular Biology, Pathobiology and Genetics

Candidate Genes in Breast Cancer Revealed by Microarray-Based Comparative Genomic Hybridization of Archived Tissue

Michelle Nessling1, Karsten Richter1, Carsten Schwaenen3, Peter Roerig4, Gunnar Wrobel1,5, Swen Wessendorf3, Björn Fritz1, Martin Bentz3, Hans-Peter Sinn2, Bernhard Radlwimmer1 and Peter Lichter1

1 Abteilung Molekulare Genetik, Deutsches Krebsforschungszentrum and 2 Pathologisches Institut, Universität Heidelberg, Heidelberg, Germany; 3 Abteilung Innere Medizin III, Medizinische Universitätsklinik und Poliklinik, Ulm, Germany; 4 Institut für Neuropathologie, Heinrich-Heine-Universität, Düsseldorf, Germany; and 5 Abteilung Bioinformatik und Biochemie, Schweizer Institut für Bioinformatik, Basel, Switzerland

Requests for reprints: Peter Lichter, Abteilung Molekulare Genetik, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 580, D-69120 Heidelberg, Germany. E-mail: m.macleod{at}dkfz.de.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results and Discussion
 References
 
Genomic imbalances in 31 formalin-fixed and paraffin-embedded primary tumors of advanced breast cancer were analyzed by microarray-based comparative genomic hybridization (matrix-CGH). A DNA chip was designed comprising 422 mapped genomic sequences including 47 proto-oncogenes, 15 tumor suppressor genes, as well as frequently imbalanced chromosomal regions. Analysis of the data was challenging due to the impaired quality of DNA prepared from paraffin-embedded samples. Nevertheless, using a method for the statistical evaluation of the balanced state for each individual experiment, we were able to reveal imbalances with high significance, which were in good concordance with previous data collected by chromosomal CGH from the same patients. Owing to the improved resolution of matrix-CGH, genomic imbalances could be narrowed down to the level of individual bacterial artificial chromosome and P1-derived artificial chromosome clones. On average 37 gains and 13 losses per tumor cell genome were scored. Gains in more than 30% of the cases were found on 1p, 1q, 6p, 7p, 8q, 9q, 11q, 12q, 17p, 17q, 20q, and 22q, and losses on 6q, 9p, 11q, and 17p. Of the 51 chromosomal regions found amplified by matrix-CGH, only 12 had been identified by chromosomal CGH. Within these 51 amplicons, genome database information defined 112 candidate genes, 44 of which were validated by either PCR amplification of sequence tag sites or DNA sequence analysis.

Key Words: breast cancer • matrix-CGH • array-CGH • microarray • candidate genes


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results and Discussion
 References
 
Breast cancer is the most frequent cancer in females worldwide. One salient feature of breast cancer is multiple genetic alterations, which correlates with its complex biology and clinical behavior. Because chromosomal imbalances correlate with altered gene expression (1), their detection is crucial in the search for relevant genes in tumorigenesis, notably proto-oncogenes and tumor suppressor genes. In the past, cytogenetic and molecular studies on breast cancer had identified frequently imbalanced chromosomal regions including loss on 1p, 4p, 4q, 8p, 11q, 13q, 16q, 17p, 18q, and 22q and gain on 1q, 8q, 11q, 12q, 16p, 17q, and 20q (2–9), leading to the characterization of several candidate genes, for example, MYC at 8q24; CCND1, EMS1, and PRAD1 at 11q13; MDM2 at 12q13; ERBB2 at 17q21;RPS6KB1, MUL, APPBP2, TRAP240, RAD51C, and BCAS3 at 17q23; and AIB1, BTAK, NABC1, ZNF217, and BCAS4 at 20q11-q13 (10–21). Here we use microarray-based comparative genomic hybridization for high-resolution analysis of genomic imbalances, an approach termed "matrix-CGH" (22) or "array-CGH" (23). Similar to chromosomal CGH (24), matrix-CGH compares the abundance of specific genomic sequences in whole tumor DNA relative to normal reference genomes. The specific test sequences, represented by DNA of well-characterized bacterial artificial chromosome (BAC) and P1-derived artificial chromosome (PAC) clones, are spotted on glass slides (microarray) in which the differently fluorescent-labeled tumor and reference DNAs are cohybridized to. Genomic imbalances are then indicated by a shift of the fluorescence signal ratios compared with balanced clones on the hybridized chip.

Matrix-CGH allows detection of imbalances below single-gene level (25). In contrast, chromosomal CGH is limited to the banding resolution of metaphase chromosomes, that is, to roughly 10 Mbp (26–28).

Typically, genomic microarrays are designed for specific purposes, for example, arrays of known proto-oncogenes and tumor suppressor genes, disease-specific arrays (29), arrays of physically ordered contiguous DNA fragments (30), and whole genome arrays with 1-Mbp resolution (31–34) or even tiling resolution (35). For the current study, a dedicated DNA microarray was generated with 422 genomic sequences of the human autosomes. This microarray comprised 222 test sequences of interest for tumor relevance and 200 test sequences representing the entire human genome at 15-Mbp resolution. Archived formalin-fixed and paraffin-embedded breast tumor tissue was analyzed from 31 patients, for which chromosomal CGH data were available.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results and Discussion
 References
 
Tumor Specimens. Whole genomic DNA was isolated according to a standard protocol (36) from formalin-fixed and paraffin-embedded tissues of primary carcinoma from 32 female patients with advanced breast cancer and involvement of at least 10 axillary lymph nodes. All test DNAs had been analyzed with CGH to metaphase chromosomes previously (partly in ref. 8). Reference DNA was isolated from peripheral blood lymphocytes and pooled from three apparently healthy donors.

DNA Microarray. The DNA microarray comprised 422 mapped sequences of human autosomes and 49 mapped sequences of the X chromosome (for chromosomal localization, name, and identification number see Supplementary Table 5 online). Spaced roughly every 15 Mbp, 200 sequences represent the entire human genome. The residual 222 DNA sequences were selected for both chromosomal regions recurrently imbalanced in cancer cells and specific genes of oncogenic potential, either screening the National Center for Biotechnology Information database or by BLAST search against finished and unfinished genome sequence data (http://www.ncbi.nlm.nih.gov). The selected material was verified for 47 oncogenes and 15 tumor suppressor genes by PCR amplification of sequence tag sites (STS-PCR) or DNA sequence analysis.

DNA sequences from BAC library RPCIB753 (equal to RP11) and PAC library RPCIP704 (equal to RP1, 3, 4, and 5) as well as additional gene-specific sequences, which were identified by filter hybridization of BAC and PAC libraries (RZPD, Berlin, Germany) using selected cDNAs as probes (25), were ordered from the German Resource Center (RZPD). Genomic BAC or PAC DNA was isolated from bacterial cultures according to standard protocols (Qiagen, Hilden, Germany). DNAs were sheared to fragment sizes of 500 to 5,000 bp (23, 25) with a final concentration of 0.5 to 1 µg/µL DNA in 3x SSC. Spotting onto CMT-GAPSll coated glass slides (Corning, Wiesbaden, Germany) was done using an Omnigrid microarrayer (GeneMachines, San Carlos, CA) equipped with SMP3 split pins. Each test sequence was spotted in eight replicas. After spotting, the slides were baked for 10 minutes at 80°C, the DNA was cross-linked by UV light of 254 nm, and the slides were stored at room temperature until use.

Matrix-CGH. In matrix-CGH, the differential sequence composition of tumor DNA compared with wild-type DNA is monitored by the ratio of hybridization efficiency to the target DNA spotted on glass slides. That for tumor and control DNA, labeled with different fluorescent dyes, are cohybridized. In the present study, CGH was done as a "color-switch" experiment, that is, each experiment comprised two cohybridizations on two different slides, one with tumor DNA labeled by Cy3 and the control DNA labeled by Cy5, and the other labeled vice versa.

Cy3- and Cy5-conjugated dCTPs (Perkin-Elmer Life Science Products, Boston, MA) were incorporated by random primer extension (BioPrime DNA labeling kit, Invitrogen, Karlsruhe, Germany) with 100 to 300 ng genomic DNA as template. Unincorporated nucleotides were removed by membrane column centrifugation (Microcon YM-30, Millipore, Billerica, MA) before the dye incorporation rates were determined spectrophotometrically. Equal amounts (3 to 4 µg) of labeled reference and tumor DNA were coprecipitated with 75 µg of human Cot-1 DNA (Invitrogen) and resolved in 120 µL prewarmed hybridization buffer (ULTRAhyb buffer, Ambion, Austin, TX). After denaturation (10 minutes at 75°C) and preannealing (1 hour at 37°C), hybridization was allowed for 42 hours at 37°C in an automated hybridization chamber (GeneTac, Genomic Solutions, Huntingdon, Cambridgeshire, United Kingdom). After automated posthybridization washes in the hybridization chamber [30 seconds "flow" in 50% FA/2x SSC/0.1% Tween 20 (pH 7) at 37°C and thrice for 20 seconds flow and 3 minutes "hold" at 45°C with the same buffer, followed by washing for 2 minutes flow in 1x PBS/0.05% Tween 20 (pH 7) at 25°C], slides were spin-dried for 5 minutes at 900 x g.

Data Acquisition and Evaluation. The hybridized microarrays were scanned with a GenePix 4000 A (Axon Instruments, Foster City, CA) and signal processing was done with the imaging software GenePixPro 3.0. The raw data comprised four images representing the two pairs of Cy3 and Cy5 channels of a color-switch experiment. First, spatial shift between the two channels of a Cy3/Cy5 pair was corrected and the spot areas were delineated interactively. Then, the data set was cleared of data of bad quality, for example, spots with too high background due to local contamination, spots in which the mean intensity was close to background in one of the two channels, and spots with compromised definition due to merge with a neighboring spot. This quality refinement rejected roughly 8% of the test sequences on the microarray. From the cleared data set, spot intensities were corrected by the local background and ratios of the median were calculated. Then, from each set of replicates the spots with the highest and the lowest ratio were eliminated. Of the remaining, maximally six and at least three replicates per test sequence, the mean of the ratios was calculated, each color switch pair combined (division and root extraction) and further analyzed with spreadsheet software (Microsoft Excel).

Identifying imbalanced genomic sequences by their CGH ratio at a predefined probability level requires knowledge of the frequency distribution of the ratios from balanced genomic sequences in the same experiment. It was shown in control experiments that the frequency distribution of the log2-transformed ratios from balanced genomic sequences is well described by a Gaussian. For uncharacterized material from a patient, there are no means to determine a priori the balanced portion of the test DNA. Thus, a reasonable guess for a pool of balanced test sequences (i.e., a "control cluster") will include some unknown number of sequences, which are not balanced. To account for those "false-negative" values when estimating the mean and SD for the balanced cases, we chose to fit a Gaussian to the major peak of the log2-transformed ratio distribution from our control cluster instead of arithmetically calculating mean and deviation over all values of the cluster. The mean of the fitted Gaussian then was subtracted from all log2-transformed ratios of the data set for normalization. The SD of the Gaussian was used to set a threshold for scoring imbalances, which is to refuse the null hypothesis: the given ratio is compatible with the balanced state. Test sequences were scored as imbalanced when their log2-transformed ratios were higher than either thrice the SD for individual measurements, or 1.9 times the SD for genomic sequences in a 15-Mbp neighborhood. Thus, all experiments were interpreted on the base of a constant confidence level of 99.8% for the null hypothesis instead of a constant threshold. Therefore, fewer imbalances are scored in experiments with higher statistical noise. This is conspicuous on the tables in which the patients were ordered with increasing noise from 1 to 31.

A control cluster was defined by an iterative procedure: As the first step, the total data set of log2-transformed ratios was normalized to its arithmetic mean. Then, a profile was plotted of the ratios against the physical neighborhood of all genomic sequences (i.e., the X axis represented the test sequences in chromosomal order). Applying a "neighborhood argument," which states that the log2-transformed ratios of balanced genomic sequences in direct neighborhood should statistically scatter around zero value, this plot was searched for contiguous chromosomal regions characterized by sign changes within a neighborhood of 30 Mbp, and the corresponding test sequences were collected into a control cluster. The histogram of log2-transformed ratios of this control cluster was then analyzed and the refined normalization reevaluated. For curve fitting, we used the software KaleidaGraph (Synergy Software, Reading, PA). For the binning, to create a histogram the square root of the number of test sequences in the control cluster was chosen as the number of bins. The program uses the method of Marquardt and Levenberg for the fitting procedure. Quality of the fit was evaluated by Pearson's R. One of the 32 experiments was declined because the R value of the fit did not reach 0.96.

Grade of Imbalance. The measured CGH ratio is not a direct measure for the copy number imbalance of the test genome in question because tumor material from a patient normally also contains an unknown fraction of healthy, wild-type cells (e.g., from cells of the connective tissue). The higher the fraction of this "intermixture," the smaller becomes the difference of an imbalanced ratio from the balanced state. In high-throughput experiments screening many tumors, the intermixture is normally not assessed independently by counting the relation of tumor nuclei versus wild-type nuclei within the material finally used for DNA extraction. In practice, it is often agreed to score imbalances by fixed thresholds (e.g., 0.75 for loss, 1.25 for gain, and 2.0 for higher-level amplification; refs. 26, 37, 38 ). In contrast, in this study we exploited the periodicity of ratios of imbalanced clones to evaluate the "grade of imbalance." In contrast to the mere ratio value, the periodicity number intrinsically considers the intermixture. Eq. A shows that for constant intermixture, f, the linear ratios scatter periodically.

(A)
where r is the linear ratio; pi, the copy number; 2, the biallelic copy number of control cells; f, the fraction of intermixture; {Delta}r, the period between matrix-CGH ratios of pi, and pi + 1 is independent of p.

Thus, once the period {Delta}r for a given experiment is known (i.e., the ratio increase caused by one increase of copy number), the scale of ratio can be divided into bins of increasing "apparent copy number." For the results presented here, the periodicities were assessed visually on the frequency distribution of the linear ratios of all nonbalanced genomic sequences from a patient. The periods had to fulfill the following conditions: they must be smaller than 0.5, only two periods are allowed in the loss regime, and they must be small enough not to create balanced scores in that part of the data that has been determined nonbalanced by the Gauss-fit procedure at a confidence level of 99.8%. The approach was verified with material from culture cells (HL60) of known imbalances.

The determined period was then used to split the scale of linear ratios into bins, one bin for each apparent copy number. The bin boundaries were calculated with Eq. B, which takes into account that the noise, x, of ratios in linear scale does not scatter symmetrically:

(B)

The identity refers to the intersection, in which clones with copy number pi adopt the same ratio as clones with copy number pi + 1 due to a relative error x/p.

Instead of apparent copy number, p, in the presented tables we list the "grade of imbalance," n = (p – 2), which is zero for balanced clones, negative for losses and positive for gains.

Although the grade of imbalance is a good estimator, the significance of its absolute value decreases rapidly with increasing n. Furthermore, with an initial screening for aberrations of an uncharacterized tumor one cannot consider a possible heterogeneous composition by different tumor stages, which would translate into a variability of intermixture among clones. The periodicity among imbalances of a heterogeneous tumor bears on the mass-dominant fraction of aberrations, whereas aberrations that only occur in a subpopulation of tumor cells will not fit well into this periodic pattern. Therefore, care must be taken to interpret the grade of imbalance in terms of real copy number. For the final interpretation of the data, the measured imbalances were scored into one of three categories: (a) loss (total loss, n = –2, and single loss, n = –1), (b) low-level gain (n = 1 and n = 2), and (c) high-level gain (n > 2).


   Results and Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results and Discussion
References
 
Comparison of Chromosomal and Matrix-CGH Data. All test DNAs were analyzed with both methods, chromosomal CGH (8) and matrix-CGH (this study). The good concordance of the results corroborates the method of data analysis that we elaborated for the interpretation of matrix-CGH experiments. In particular, evaluation of the grade of imbalance seemed more consistent within neighborhoods and among patients (see Table 1, neighborhood, e.g., in 1p36, 1q, 7p12, 8q24, or 12q) than on application of fixed thresholds (data not shown).


View this table:
[in this window]
[in a new window]

  		Table 1. Genomic imbalances found by matrix-CGH of 31 cases of advanced breast cancer

 
Although matrix-CGH mirrors the previous results of chromosomal CGH, the information about genomic imbalances is more detailed due to improved resolution with respect to the physical size of the aberrant region and the grade of imbalance (supporting information in Supplementary Table 4 online). The amplification within 12q14, which was scored in 19 cases (61%) by matrix-CGH, had been detected in only 2 cases by chromosomal CGH. In these 2 cases, a region of 70 Mbp was affected, whereas in the other 17 cases the amplification measured by matrix-CGH spanned only 2 Mbp. Similarly, the amplifications within regions of low-level gain in 11q13 were well represented by matrix-CGH but had not shown up with chromosomal CGH. The neighborhood of repetitive genomic regions are known to often lead to false-positive results in chromosomal CGH. This problem does not exist in matrix-CGH in which each test sequence is spatially well separated. Accordingly, high-level gain on 7p12, adjacent to the centromeric region, had not been assessed previously by chromosomal CGH (Fig. 1).



View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Comparison of matrix-CGH ratios for chromosome 7 of case 2 with chromosomal CGH of the same case in a log2-transformed profile plot. Y axis, scaled according to the chromosomal ideogram (inset), with p arm, centromer (cen), and q arm. +, matrix-CGH measurements. Solid line, chromosomal CGH profile from the inset, transformed to logarithmic scale. By matrix-CGH, the amplified region at 7p11 is much sharper delimited and the measured ratios are much higher. High-level gain of the gene EGFR, coding for the epidermal growth factor receptor, was identified with six gene-specific DNA fragments, representing significant log2-transformed ratios.

 
Genomic Aberrations Detected by Matrix-CGH in Advanced Breast Cancer. Matrix-CGH of formalin-fixed and paraffin-embedded tissues from 31 cases of advanced breast cancer revealed a high number of imbalances at high resolution all over the genome. Genomic regions that were found aberrant in more than one third of the analyzed tumor cases are presented in Table 1, whereas all collected chromosomal gains and losses are summarized in Supplementary Table 4 (online). Of the 1,865 aberrant ratios detected, the majority (1,154) scored as low-level gain, 315 events were high-level gains, whereas the remaining (396) were assessed as loss. Biallelic loss, a rare event in solid tumors, was detected sporadically in 12 cases (39%). Remarkably, total loss of the single clone RP11-2M9 at 6q12 was found in 6 cases (19%), pointing to a tumor suppressor gene locus. Although aberrations were found all over the genome, some hotspots of imbalance could be identified: loss in at least 25% of the analyzed cases was observed on chromosomal subregions 6q12, 9p21, 9p22-p23, and 11q22-q23; low- and high-level gain in at least 45% of the cases was found in chromosomal subregions 1p36, 1q, 3q26, 6p21, 8q24, 11q13, 12q13-q15, 17q21-q22, and 20q11-q13 (Table 2).


View this table:
[in this window]
[in a new window]

  		Table 2. Incidence of low- and high-level imbalances in advanced breast cancer (n = 31)

 
With respect to high-level gain, 51 genomic regions were identified. The respective BACs or PACs were tested positive for 44 specific genes either by STS-PCR or by DNA sequence analysis. According to genome database information, further 68 genes are included as candidates (Table 3).


View this table:
[in this window]
[in a new window]

  		Table 3. Incidence of amplified candidate genes identified by matrix CGH in 31 cases of advanced breast cancer

 
Genomic imbalances observed with high incidence in a well-defined chromosomal region are indicative of candidate genes in tumor development. In the following, a selection of chromosomal regions with highest frequencies of aberration is discussed in more detail (Table 1).

Chromosomal Regions Frequently Affected in Breast Cancer. On 7p12, the EGFR/ERBB1 gene was tested with seven different clones. Gain within this region was observed in 11 cases (35%); in one case, high-level gain was detected in six of these seven test sequences. The epidermal growth factor receptor (EGFR/ERBB1) belongs to the gene family of four ERBB/HER protein-tyrosine kinases, of which ERBB1 and ERBB2 (17q21) are overexpressed in roughly one third of breast carcinomas (39). Whereas ERBB2 overexpression is known to correlate with gene amplification (see below), ERBB1 expression seems to be regulated in different cellular levels: According to Kersting et al. (40), only 55% of patients with genomic gain also exhibited increased gene expression. On the other hand, patients with no gain detected displayed increased gene expression in 13% of the cases. Thus, whereas amplification correlates with overexpression of the gene, it is neither necessary nor sufficient for increased expression levels. In the present study, the incidence of gain (35%) was higher compared with other studies in the literature (0-15%, refs. 41, 42). Possibly, the higher incidence of EGFR/ERBB1 gain depends on the advanced stage of all tumors investigated.

The region around the MYC oncogene locus on 8q24 was included in the study by a well-characterized contig of 10 genomic PAC clones. Similar to published data (43, 44), the MYC gene was amplified in 5 cases (16%). Roughly half of the analyzed cases (48%) were affected by low-level gain.

Loss on 9p was detected in 17 cases (55%). In concordance with Kamb et al. (45) the BAC clone RP11-149I2 on 9p21 was tested positive for the tumor suppressor genes CDKN2A (p16) and CDKN2B (p15). Both cyclin-dependent kinase inhibitors are frequently modified in various types of tumors, including breast cancer. In this context it seems noteworthy that female carriers of mutations in the melanoma susceptibility gene, CDKN2A, exhibit a higher than expected risk of breast cancer (46, 47).

11q frequently was affected by both gain and loss, with loss in the distal part, in particular on 11q22-q23. The lost region includes the locus for the tumor suppressor gene ATM, which was explicitly tested for imbalance by a PAC clone verified for ATM. Both DNA double-strand-break checkpoint genes, ATM on 11q22-q23 and BRCA1 on 17q21, respectively, and the repair gene TP53 on 17p13, act together in a functional pathway (48): ATM serves as the upstream sensor, which on DNA double-strand-break phosphorylates BRCA1 and TP53 to trigger repair and cell cycle regulation. Frequent deletion of these three gene loci in sporadic breast cancer is discussed (49–51). In line with this concept, in the present study codeletion of distal 11q and 17p was observed in 6 cases (19%), whereas in 1 case (case 2) even all three chromosomal regions were affected.

Gain in 11q12-q13 occurred in 24 cases (77%), roughly one half of them with amplification. The affected genes were CCND1/PRAD1/BCL1, FGF3/INT2, FGF4/HSTF1, RARRES3, RELA, and BAD. Of the amplified genes, CCND1, promoting cell cycle transit, is of main interest, because Cuny et al. (17) described that amplification of CCND1 is associated with shorter overall survival in patients with breast cancer. However, a pathogenic role of the other candidate genes should not be excluded.

The well known hotspot on 17q harbors several breast cancer–related genes. In our study, 26 cases (84%) showed gain, 12 (39%) of which at high level. Within 17q12-q21, a minimal region of roughly 1 Mbp, represented by seven genomic sequences, was most frequently involved. In addition to RARA and TOP2A, this region includes the oncogene ERBB2/HER2/NEU, in which increased gene expression correlates with gene amplification, in contrast to ERBB1 on 7p (52). ERBB2/HER2/NEU overexpression is a predictor for poor clinical outcome.

Novel Findings of Genomic Imbalance in Advanced Breast Cancer. Chromosome 1 was frequently affected by gains on the p arm as well as on the q arm. In 1p36, represented by 11 BAC and PAC clones, amplifications are interrupted by a gap of balanced genomic sequences: The distal part of gain, seen in 19 cases (60%) and including the genes TP73 and TNFRSF12, comprised roughly 6 Mbp, followed by four genomic clones within a region of 17 Mbp in which only two aberrations were found, and the next two genomic clones, representative of the genes E2F2 and FGR, again gained at a high incidence. Although Dominguez et al. (53) showed a statistical trend of poor prognosis for breast cancer with high p73 level, imbalances on the gene-rich subregion 1p36 have not been considered yet to be of major relevance in tumorigenesis of the breast.

Gain on 6p21 was identified in 14 cases (45%). The region, represented by 11 test sequences, included five genomic clones selected for CCND3 and one genomic clone for CDKN1A/p21/WAF1. Both candidate genes had not been described as frequently imbalanced in tumor cell genomes of the breast.

The protein encoded by CCND3, gained in one third of the analyzed cases, belongs to the family of highly conserved cyclins, regulators of cyclin-dependent kinases (CDK). CCND3 functions as a regulatory subunit of CDK4 or CDK6 in G1-S transition. The protein encoded by CDKN1A, overrepresented in 7 cases (23%), inhibits cyclin kinase activity and probably serves as the effector of p53-mediated cell cycle control.

Gain of 12q was seen in 19 of the analyzed cases (61%). Most of the genomic sequences tested represent the subregion 12q13-q15, including the gene loci GLI (12q13.3), SAS and CDK4 (both 12q14.1), TIP120A (12q14.3), and MDM2 (12q15), all verified by STS-PCR.

This genomic region is frequently amplified in various tumors. However, it is normally not assigned as a hotspot of imbalance in breast cancer like 8q, 11q, 17q, or 20q. One reason may be that the low-level gain is usually ignored when fixed thresholds are applied to score imbalances in tumor material with high fractions of intermixture.

The highly recurrent alteration of 20q, observed in 22 cases (71%), concerns genomic clones representing AIB1/NCOA3 (20q12), involved in all of these cases, in one third of them with high-level gain. The nuclear receptor coactivator AIB1 (Amplified in Breast cancer 1) interacts with estrogen receptors and enhances estrogen-dependent transcription. Anzick et al. (11) suggests that altered AIB1 expression contributes to development of steroid-dependent cancers. Accordingly, we see a 100% correlation (8 cases, 26%) between gain for clone RP3-443C4 (on 6q25), positively tested for ESR1 (estrogen receptor), and gain of AIB1/NCOA3.

The adjacent region, 20q13.1, represented by four test sequences, was gained in 18 of the analyzed cases (58%). Among them, one third exhibiting high-level gain on clone RP4-530I15 positively tested for MYBL2 and ZNF217. Whereas MYBL2 is a nuclear protein involved in cell cycle progression, a specific function for MYBL2 in breast cancer has not been described.

The most distal genomic clone on 20q13 (RP5-1167H4), representing STK6/AURORA-A, was overrepresented in 10 cases (32%). Because the protein encoded by this gene is a cell cycle–regulated kinase, which seems to be involved in microtubule formation and/or stabilization at the spindle pole during chromosome segregation, STK6/AURORA-A may play a role in tumor development and progression by inducing mitotic chromosomal segregation (54).

Outlook. Both chromosomal and matrix-CGH allow mapping of DNA sequence copy number variations in tumor tissues. The resolution limit of chromosomal CGH is within 10 Mbp, whereas matrix-CGH is able to detect imbalances at the level of single genes. Thus, the limiting factor for the detection of imbalances by matrix-CGH is rather the quality of the tumor DNA than the size of the affected region. As a standard in clinical pathology, specimens are often formalin fixed and paraffin embedded, which compromises the prepared DNA due to degradation, causing a great variability in statistical noise among individual matrix-CGH experiments. Therefore, we aimed to analyze the data based on constant statistical significance instead of using fixed thresholds for the scoring of imbalances. In experiments with low noise, we detected even low-level imbalances with high statistical significance, showing that matrix-CGH in combination with elaborate statistical evaluation of each individual experiment is a powerful tool for high-resolution screening of chromosomal imbalances in archived material from breast cancer tissues. We further propose to exploit the periodicity of ratios of imbalanced clones for evaluation of the grade of imbalance. In contrast to the mere ratio value, this approach intrinsically considers the intermixture, thus allowing us to compare imbalances among patients more directly. As we note in MATERIALS AND METHODS, the statistical significance of the grade of imbalance is not well defined and decreases with increasing value. Furthermore, because tumor heterogeneity was not considered, the value should be taken as a good estimator rather than a direct measure for the real copy number deviation. These two steps in data analysis, that is, evaluation of all individual experiments to a constant level of significance and the transformation of ratio values into grades of imbalance, substantially improved the consistency within the detected genomic aberrations from the archived material of 31 human breast tumors (see Supplementary Table 4).


   Acknowledgments
 
Grant support: Bundesministerium für Bildung und Forschung Germany and Nationales Genomforschungsnetz grants 01SF9909/0 and 01GR0101.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Andrea Wittmann, Daniel Göttel, Stefanie Hofmann, Felix Kokocinski, Antoaneta Mincheva, Tatjana Salvi, and Boris Zielinski (Heidelberg) and Martina Enz and Holger Kohlhammer (Ulm) for excellent technical assistance and sharing of material.


   Footnotes
 
Note: Supplementary data for this are available at Cancer Research Online (http://cancerres.aacrjournal.org).

Received 8/20/04. Revised 10/25/04. Accepted 11/12/04.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results and Discussion
 References
 

  1. Gene BO. Amplification and tumor progression. Biochim Biophys Acta 1993;1155:25–41.[Medline]
  2. Kallioniemi A, Kallioniemi OP, Piper J, et al. Detection and mapping of amplified DNA sequences in breast cancer by comparative genomic hybridization. Proc Natl Acad Sci U S A 1994;91:2156–60.[Abstract/Free Full Text]
  3. Isola JJ, Kallioniemi OP, Chu LW, et al. Genetic aberrations detected by comparative genomic hybridization predict outcome in node-negative breast cancer. Am J Pathol 1995;147:905–11.[Abstract]
  4. Ried T, Just KE, Holtgreve-Grez H, et al. Comparative genomic hybridization of formalin-fixed, paraffin-embedded breast tumors reveals different patterns of chromosomal gains and losses in fibroadenomas and diploid and aneuploid carcinomas. Cancer Res 1995;55:5415–23.[Abstract/Free Full Text]
  5. Zitzelsberger H, Lehmann L, Werner M, Bauchinger M. Comparative genomic hybridisation for the analysis of chromosomal imbalances in solid tumours and haematological malignancies. Histochem Cell Biol 1997;108:403–17.[CrossRef][Medline]
  6. Knuutila S, Bjorkqvist AM, Autio K, et al. DNA copy number amplifications in human neoplasms: review of comparative genomic hybridization studies. Am J Pathol 1998;152:1107–23.[Abstract]
  7. Richard F, Pacyna-Gengelbach M, Schluns K, et al. Patterns of chromosomal imbalances in invasive breast cancer. Int J Cancer 2000;89:305–10.[CrossRef][Medline]
  8. Seute A, Sinn HP, Schlenk RF, et al. Clinical relevance of genomic aberrations in homogeneously treated high-risk stage II/III breast cancer patients. Int J Cancer 2001;93:80–4.[CrossRef][Medline]
  9. Climent J, Martinez-Climent JA, Blesa D, et al. Genomic loss of 18p predicts an adverse clinical outcome in patients with high-risk breast cancer. Clin Cancer Res 2002;8:3863–9.[Abstract/Free Full Text]
  10. Schuuring E, Verhoeven E, Mooi WJ, Michalides RJ. Identification and cloning of two overexpressed genes, U21B31/PRAD1 and EMS1, within the amplified chromosome 11q13 region in human carcinomas. Oncogene 1992;7:355–61.[Medline]
  11. Anzick SL, Kononen J, Walker RL, et al. AIB1, a steroid receptor coactivator amplified in breast and ovarian cancer. Science 1997;277:965–8.[Abstract/Free Full Text]
  12. Sen S, Zhou H, White RA. A putative serine/threonine kinase encoding gene BTAK on chromosome 20q12 is amplified and overexpressed in human breast cancer cell lines. Oncogene 1997;14:2195–200.[CrossRef][Medline]
  13. Collins C, Rommens JM, Kowbel D, et al. Positional cloning of ZNF217 and NABC1: genes amplified at 20q13.2 and overexpressed in breast carcinoma. Proc Natl Acad Sci U S A 1998;95:8703–8.[Abstract/Free Full Text]
  14. Couch FJ, Wang XY, Wu GJ, Qian J, Jenkins RB, James CD. Localization of PS6K to chromosomal region 17q23 and determination of its amplification in breast cancer. Cancer Res 1999;59:1408–11.[Abstract/Free Full Text]
  15. Bärlund M, Forozan F, Kononen J, et al. Detecting activation of ribosomal protein S6 kinase by complementary DNA and tissue microarray analysis. J Natl Cancer Inst 2000;92:1252–9.[Abstract/Free Full Text]
  16. Bärlund M, Monni O, Kononen J, et al. Multiple genes at 17q23 undergo amplification and overexpression in breast cancer. Cancer Res 2000;60:5340–4.[Abstract/Free Full Text]
  17. Cuny M, Kramar A, Courjal F, et al. Relating genotype and phenotype in breast cancer: an analysis of the prognostic significance of amplification at eight different genes or loci and of p53 mutations. Cancer Res 2000;60:1077–83.[Abstract/Free Full Text]
  18. Wu GJ, Sinclair CS, Paape J, et al. 17q23 amplifications in breast cancer involve the PAT1, RAD51C, PS6K, and SIGMA1B. Cancer Res 2000;60:5371–5.[Abstract/Free Full Text]
  19. Wu GJ, Sinclair C, Hinson S, Ingle JN, Roche PC, Couch FJ. Structural analysis of the 17q22-q23 amplicon identifies several independent targets of amplification in breast cancer cell lines and tumors. Cancer Res 2001;61:4951–5.[Abstract/Free Full Text]
  20. Monni O, Bärlund M, Mousses S, et al. Comprehensive copy number and gene expression profiling of the 17q23 amplicon in human breast cancer. Proc Natl Acad Sci U S A 2001;98:5711–6.[Abstract/Free Full Text]
  21. Bärlund M, Monni O, Weaver JD, et al. Cloning of BCAS3 (17q23) and BCAS4 (20q13) genes that undergo amplification, overexpression, and fusion in breast cancer. Genes Chromosomes Cancer 2002;35:311–7.[CrossRef][Medline]
  22. Solinas-Toldo S, Lampel S, Benner A, Cremer T, Lichter P. Matrix-based comparative genomic hybridization: biochips to screen for genomic imbalances. Genes Chromosomes Cancer 1997;20:399–407.[CrossRef][Medline]
  23. Pinkel D, Segraves R, Sudar D, et al. High resolution analysis of DNA copy number variation using comparative genomic hybridization to microarrays. Nat Genet 1998;20:207–11.[CrossRef][Medline]
  24. Kallioniemi A, Kallioniemi OP, Sudar D, et al. Comparative genomic hybridization for molecular cytogenetic analysis of solid tumors. Science 1992;258:818–21.[Abstract/Free Full Text]
  25. Wessendorf S, Fritz B, Wrobel G, et al. Automated screening for genomic imbalances using matrix-based comparative genomic hybridization. Lab Invest 2002;82:47–60.[Medline]
  26. Kallioniemi OP, Kallioniemi A, Piper J, et al. Optimizing comparative genomic hybridization for analysis of DNA sequence copy number changes in solid tumors. Genes Chromosomes Cancer 1994;10:231–43.[Medline]
  27. Bentz M, Plesch A, Stilgenbauer S, Döhner H, Lichter P. Minimal sizes of deletions detected by comparative genomic hybridization. Genes Chromosomes Cancer 1998;21:172–5.[CrossRef][Medline]
  28. Lichter P, Joos S, Bentz M, Lampel S. Comparative genomic hybridization: uses and limitations. Semin Hematol 2000;37:348–57.[CrossRef][Medline]
  29. Schwaenen C, Nessling M, Wessendorf S, et al. Automated array-based genomic profiling in chronic lymphocytic leukemia: development of a clinical tool and discovery of recurrent genomic alterations. Proc Natl Acad Sci U S A 2004;101:1039–44.[Abstract/Free Full Text]
  30. Albertson DG, Ylstra B, Segraves R, et al. Quantitative mapping of amplicon structure by array CGH identifies CYP24 as a candidate oncogene. Nat Genet 2000;25:144–6.[CrossRef][Medline]
  31. Fiegler H, Carr P, Douglas EJ, et al. DNA microarrays for comparative genomic hybridisation based on DOP-PCR amplification of BAC and PAC clones. Genes Chromosomes Cancer 2003;36:361–74.[CrossRef][Medline]
  32. Veltman JA, Fridlyand J, Pejavar S, et al. Array-based comparative genomic hybridization for genome-wide screening of DNA copy number in bladder tumors. Cancer Res 2003;63:2872–80.[Abstract/Free Full Text]
  33. Hurst CD, Fiegler H, Carr P, Williams S, Carter NP, Knowles MA. High-resolution analysis of genomic copy number alterations in bladder cancer by microarray-based comparative genomic hybridization. Oncogene 2004;23:2250–63.[CrossRef][Medline]
  34. Shaw-Smith C, Redon R, Rickman L, et al. Microarray based comparative genomic hybridisation (array-CGH) detects submicroscopic chromosomal deletions and duplications in patients with learning disability/mental retardation and dysmorphic features. J Med Genet 2004;41:241–8.[Abstract/Free Full Text]
  35. Ishkanian AS, Malloff CA, Watson SK, et al. A tiling resolution DNA microarray with complete coverage of the human genome. Nat Genet 2004;36:299–303.[CrossRef][Medline]
  36. Sambrook J, Fritsch EF, Maniatis T. In: Molecular cloning: A laboratory manual. 2nd ed. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory; 1989. p. 9.16–9.19.
  37. Bentz M, Döhner H, Huck K, et al. Comparative genomic hybridization in the investigation of myeloid leukemias. Genes Chromosomes Cancer 1995;12:193–200.[Medline]
  38. du Manoir S, Schröck E, Bentz M, et al. Quantitative analysis of comparative genomic hybridization. Cytometry 1995;19:27–41.[CrossRef][Medline]
  39. Roskoski R Jr. The ErbB/HER receptor protein-tyrosine kinases and cancer. Biochem Biophys Res Commun 2004;319:1–11.[CrossRef][Medline]
  40. Kersting C, Tidow N, Schmidt H, et al. Gene dosage PCR and fluorescence in situ hybridization reveal low frequency of egfr amplifications despite protein overexpression in invasive breast carcinoma. Lab Invest 2004;84:582–7.[CrossRef][Medline]
  41. Ro J, North SM, Gallick GE, Hortobagyi GN, Gutterman JU, Blick M. Amplified and overexpressed epidermal growth factor receptor gene in uncultured primary human breast carcinoma. Cancer Res 1988;48:161–4.[Abstract/Free Full Text]
  42. Guérin M, Gabillot M, Mathieu MC, et al. Structure and expression of c-erbB-2 and EGF receptor genes in inflammatory and non-inflammatory breast cancer: prognostic significance. Int J Cancer 1989;43:201–8.[Medline]
  43. Deming SL, Nass SJ, Dickson RB, Trock BJ. C-myc amplification in breast cancer: a meta-analysis of its occurrence and prognostic relevance. Br J Cancer 2000;83:1688–95.[CrossRef][Medline]
  44. Rummukainen JK, Salminen T, Lundin J, Kytola S, Joensuu H, Isola JJ. Amplification of c-myc by fluorescence in situ hybridization in a population-based breast cancer tissue array. Mod Pathol 2001;14:1030–5.[CrossRef][Medline]
  45. Kamb A, Gruis NA, Weaver-Feldhaus J, et al. A cell cycle regulator potentially involved in genesis of many tumor types. Science 1994;8:23–6.
  46. Borg A, Sandberg T, Nilsson K, et al. High frequency of multiple melanomas and breast and pancreas carcinomas in CDKN2A mutation-positive melanoma families. J Natl Cancer Inst 2000;92:1260–6.[Abstract/Free Full Text]
  47. Goggins W, Gao W, Tsao H. Association between female breast cancer and cutaneous melanoma. Int J Cancer 2004;111:792–4.[CrossRef][Medline]
  48. Ding SL, Sheu LF, Yu JC, et al. Abnormality of the DNA double-strand-break checkpoint/repair genes, ATM, BRCA1 and TP53, in breast cancer is related to tumour grade. Br J Cancer 2004;90:1995–2001.[CrossRef][Medline]
  49. Lo YL, Yu JC, Huang CS, et al. Allelic loss of the BRCA1 and BRCA2 genes and other regions on 17q and 13q in breast cancer among women from Taiwan (area of low incidence but early onset). Int J Cancer 1998;79:580–7.[CrossRef][Medline]
  50. Rio PG, Pernin D, Bay JO, et al. Loss of heterozygosity of BRCA1, BRCA2 and ATM genes in sporadic invasive ductal breast carcinoma. Int J Oncol 1998;13:849–53.[Medline]
  51. Shen CY, Yu JC, Lo YL, et al. Genome-wide search for loss of heterozygosity using laser capture microdissected tissue of breast carcinoma: an implication for mutator phenotype and breast cancer pathogenesis. Cancer Res 2000;60:3884–92.[Abstract/Free Full Text]
  52. Slamon DJ, Godolphin W, Jones LA, et al. Studies of the HER-2/neu proto-oncogene in human breast and ovarian cancer. Science 1989;244:707–12.[Abstract/Free Full Text]
  53. Dominguez G, Silva JM, Silva J, et al. Wild type p73 overexpression and high-grade malignancy in breast cancer. Breast Cancer Res Treat 2001;66:183–90.[CrossRef][Medline]
  54. Li D, Zhu J, Firozi PF, et al. Overexpression of oncogenic STK15/BTAK/Aurora A kinase in human pancreatic cancer. Clin Cancer Res 2003;9:991–7.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 CarcinogenesisHome page
M. Kasper, V. Jaks, M. Fiaschi, and R. Toftgard
Hedgehog signalling in breast cancer
Carcinogenesis, June 1, 2009; 30(6): 903 - 911.
[Abstract] [Full Text] [PDF]

Home page
 Cancer Res.Home page
M. Fiaschi, B. Rozell, A. Bergstrom, and R. Toftgard
Development of Mammary Tumors by Conditional Expression of GLI1
Cancer Res., June 1, 2009; 69(11): 4810 - 4817.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
E. Beauchamp, G. Bulut, O. Abaan, K. Chen, A. Merchant, W. Matsui, Y. Endo, J. S. Rubin, J. Toretsky, and A. Uren
GLI1 Is a Direct Transcriptional Target of EWS-FLI1 Oncoprotein
J. Biol. Chem., April 3, 2009; 284(14): 9074 - 9082.
[Abstract] [Full Text] [PDF]

Home page
 Cancer Epidemiol. Biomarkers Prev.Home page
S. L. Park, S.-C. Chang, L. Cai, C. Cordon-Cardo, B.-G. Ding, S. Greenland, S. K. Hussain, Q. Jiang, S. Liu, M.-L. Lu, et al.
Associations between Variants of the 8q24 Chromosome and Nine Smoking-Related Cancer Sites
Cancer Epidemiol. Biomarkers Prev., November 1, 2008; 17(11): 3193 - 3202.
[Abstract] [Full Text] [PDF]

Home page
 Cancer Epidemiol. Biomarkers Prev.Home page
V. W. Setiawan, G. Ursin, P. L. Horn-Ross, D. Van Den Berg, L. Le Marchand, B. E. Henderson, L. Bernstein, and C. A. Haiman
Germ Line Variation at 8q24 and Endometrial Cancer Risk
Cancer Epidemiol. Biomarkers Prev., October 1, 2007; 16(10): 2166 - 2168.
[Full Text] [PDF]

Home page
 BioinformaticsHome page
A. Buness, R. Kuner, M. Ruschhaupt, A. Poustka, H. Sultmann, and A. Tresch
Identification of aberrant chromosomal regions from gene expression microarray studies applied to human breast cancer
Bioinformatics, September 1, 2007; 23(17): 2273 - 2280.
[Abstract] [Full Text] [PDF]

Home page
 Proc. Natl. Acad. Sci. USAHome page
M. Lauth, A. Bergstrom, T. Shimokawa, and R. Toftgard
Inhibition of GLI-mediated transcription and tumor cell growth by small-molecule antagonists
PNAS, May 15, 2007; 104(20): 8455 - 8460.
[Abstract] [Full Text] [PDF]

Home page
 Cancer Res.Home page
F. R. Schumacher, H. S. Feigelson, D. G. Cox, C. A. Haiman, D. Albanes, J. Buring, E. E. Calle, S. J. Chanock, G. A. Colditz, W. R. Diver, et al.
A Common 8q24 Variant in Prostate and Breast Cancer from a Large Nested Case-Control Study
Cancer Res., April 1, 2007; 67(7): 2951 - 2956.
[Abstract] [Full Text] [PDF]

Home page
 Genome ResHome page
J. Hicks, A. Krasnitz, B. Lakshmi, N. E. Navin, M. Riggs, E. Leibu, D. Esposito, J. Alexander, J. Troge, V. Grubor, et al.
Novel patterns of genome rearrangement and their association with survival in breast cancer
Genome Res., December 1, 2006; 16(12): 1465 - 1479.
[Abstract] [Full Text] [PDF]

Home page
 FASEB J.Home page
J. Russo, S. V. Fernandez, P. A. Russo, R. Fernbaugh, F. S. Sheriff, H. M. Lareef, J. Garber, and I. H. Russo
17-Beta-estradiol induces transformation and tumorigenesis in human breast epithelial cells
FASEB J, August 1, 2006; 20(10): 1622 - 1634.
[Abstract] [Full Text] [PDF]

Home page
 Clin. Cancer Res.Home page
D. E. Stange, B. Radlwimmer, F. Schubert, F. Traub, A. Pich, G. Toedt, F. Mendrzyk, U. Lehmann, R. Eils, H. Kreipe, et al.
High-Resolution Genomic Profiling Reveals Association of Chromosomal Aberrations on 1q and 16p with Histologic and Genetic Subgroups of Invasive Breast Cancer
Clin. Cancer Res., January 15, 2006; 12(2): 345 - 352.
[Abstract] [Full Text] [PDF]

Home page
 Cold Spring Harb Symp Quant BiolHome page
J. HICKS, L. MUTHUSWAMY, A. KRASNITZ, N. NAVIN, M. RIGGS, V. GRUBOR, D. ESPOSITO, J. ALEXANDER, J. TROGE, M. WIGLER, et al.
High-Resolution ROMA CGH and FISH Analysis of Aneuploid and Diploid Breast Tumors
Cold Spring Harb Symp Quant Biol, January 1, 2005; 70(0): 51 - 63.
[Abstract] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Nessling, M.
Right arrow Articles by Lichter, P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nessling, M.
Right arrow Articles by Lichter, P.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online