AbstractBackground and Objectives Multiple myeloma (MM) is a malignant plasma cell neoplasia in which genetic studies have shown that genomic changes may affect almost all chromosomes, as shown by fluorescence in situ hybridization (FISH) and comparative genomic hybridization (CGH). Our objective was the genomic characterization of CD 138 positive primary MM samples by means of a high resolution array CGH platform.Design and Methods For the first time, a high resolution array CGH with more than 40,000 probes, has been used to analyze 26 primary MM samples after the enrichment of CD138-positive plasma cells.Results This approach identified copy number imbalances in all cases. Bioinformatics strategies were optimized to perform data analysis allowing the segregation of hyperdiploid and non-hyperdiploid cases by array CGH. Additional analysis showed that structural chromosome rearrangements were more frequently seen in hyperdiploid cases. We also identified the same Xq21 duplication in nearly 20% of the cases, which originated through unbalanced chromosome translocations. High level amplifications and homozygous deletions were recurrently observed in our series and involved genes with meaningful function in cancer biology.Interpretation and Conclusions High resolution array CGH allowed us to identify copy number changes in 100% of the primary MM samples. We segregated different MM subgroups based on their genomic profiles which made it possible to identify homozygous deletions and amplifications of great genetic relevance in MM.
Multiple myeloma (MM) is an incurable malignant plasma cell (PC) neoplasia characterized by the accumulation of malignant plasma cells in the bone marrow. There is a wide variability in clinical features, responses to treatment, and survival times among patients.1 Chromosome aberrations, present in virtually all patients with MM2 include chromosome translocations involving the IG loci and/or copy number changes of partial or whole chromosomes.3 Two major genetic subtypes of MM have been defined: the hyperdiploid variant (H-MM), associated with multiple chromosome trisomies, and the non-hyperdiploid variant (NH-MM) with a high prevalence of IGH translocations.4 This classification seems to have relevance in prognosis since, recently, H-MM has been associated with a better survival and a similar response rate to treatment when compared to NH-MM.5
Chromosome-based comparative genomic hybridization (CGH) has shown that DNA copy number variations (CNV) in MM may affect almost all chromosomes.6–8 We recently found, by array-based CGH (array CGH) CNV that resulted in the overexpression of genes included in amplified chromosomal regions.9 Here, we characterized CNV in a panel of CD138-enriched primary MM samples with high density array CGH.
Design and Methods
Patients and molecular cytogenetics
DNA from CD138 cells, isolated from the bone marrow of 26 cases of MM at diagnosis, by means of magnetic cell sorting (Miltenyi Biotec, Aubum, CA, USA), was studied. DNA was extracted using DNeasy tissue kits (Qiagen, Germantown, MD, USA). All samples were enriched to more than 70% plasma cells (median 95%) (Supplementary Table 1). The median age of the patients from whom the sampes were obtained was 67.5 years (range: 34–85). There were 16 men and 10 women in our series. Most patients (22/26) were homogenously treated with the GEMM2000 protocol designed by the Spanish GEM-PETHEMA group (Table 1). All samples were collected after proper informed consent. All samples were cytogenetically characterized: 85% showed a normal karyotype. We screened our series by fluorescence in situ hybridization (FISH) with IGH (Vysis, Downers Grove, IL, USA) and IGL10 break-apart probes. Nine samples (34.6%) showed IGH rearrangements and two (7.7%) had IGL rearrangements.
Reference genomic DNA
Genomic DNA was obtained from peripheral blood of ten healthy female donors. DNA was extracted using a standard saline precipitation method and was purified by phenol-chloroform extraction using Light Phase-Lock Gel tubes (Eppendorf AG, Hamburg, Germany). The same amounts of each donor DNA were mixed in a female DNA pool for use as reference DNA in all hybridizations.
Array CGH assays
Human Genome CGH 44k microarrays (Agilent Technologies, Palo Alto, CA, USA) were used in this study. This platform consists of approximately 44000 60-mer oligonucleotide probes that span the human genome with an average resolution below 50 Kb. The platform has gene focused coverage in order to ensure adequate coverage in most commonly studied genomic regions. Array CGH assays were performed as previously described11 with minor modifications. The DNA pool from ten helathy females was used as the reference DNA for all hybridizations. Slides were scanned using an Agilent 2565AA DNA Microarray Scanner (Agilent Technologies). Raw data can be freely accessed via our ftp server at www.cnio.es.
Images and data analysis
Microarray images were transformed to fluorescence intensities using Feature Extraction Software, version 8.1 (Agilent Technologies). Data were analyzed using Agilent CGH Analytics 3.2.25 software. Quality criteria were a derivative log ratio (DLR) spread lower than 0.3 log units and a signal to noise ratio for each channel greater than 30. DLR spread metrics estimates the log ratio noise by calculating the spread of log ratio differences between consecutive probes along all chromosomes. In order to establish gained and lost regions, a 500 Kb weighted moving-average window and a ζ-score threshold of ±2.5 were used. ζscore values were obtained for each probe and were categorized into gains and losses. Using this threshold, high level amplifications were defined as consecutive clones, spanning a region, which showed a log10 ratio higher than 0.3 (more than five copies) within a gained region and homozygous deletions as consecutive clones with a log10 ratio lower than −1 (no copies) and detected as losses by ζ–score. Those aberrant regions that did not reach these cut off values were considered low copy gains or single losses. For the subsequent analyses, regions of gains and losses were established. A region is defined as a group of at least three consecutive probes that are simultaneously and similarly changed.
Aberrations were categorized as numerical and structural changes. The latter type was divided into changes greater than 3Mb (>3Mb) and those less than 3Mb (<3Mb) regarding high resolution conventional CGH resolution (detailed data are provided in Supplementary Table 3).
Identification of H-MM and NH-MM
H-MM and NH-MM were easily classified in our series by array CGH. Based on the array CGH profile a copy number score (CNS) value was assigned to each chromosome. When there was not a whole chromosome gain or loss, a CNS value of 2 was assigned, when an entire chromosome gain occurred, a CNS of 3 was assigned and, finally, when a loss occurred the CNS was equal to 1. The sum of CNS values of each case was calculated and if the value was from 48 to 74 the case was classified as H-MM; when a different CNS was obtained, it was considered NH-MM. Ploidy status was further confirmed by a standard FISH assay based on centromeric probes (Vysis, Downers Grove, IL, USA).
Definition of smallest overlapping regions of imbalance (SORI)
Genomic aberration region data were used to obtain the SORI. Each SORI represented the most recurrent region of imbalance whose limits were defined by the minimum number of clones with the highest percentage of recurrence.12 SORI were defined independently for gains and losses. They were evaluated to define regions targeted by overlapping events in two or more samples. SORI limits were defined by the clones at the boundaries of the gained or lost region of the case showing the smallest unbalanced region. After SORI delimitation the presence or absence of belonging aberration was assigned to each sample (0=absence, 1=presence).
Bioinformatics and statistical analysis
GEPAS Suite software (http://gepas.bioinfo.cipf.es/) was used for unsupervised clustering, supervised analysis and gene annotation. A region was considered as statistically significant if the adjusted p value was below 0.05. For statistical comparison of the difference in the number of copy number changes in different groups, the U Mann-Whitney test was used with the SPSS v 13.0 package.
FISH and spectral karyotyping (SKY) analysis
FISH assays were carried out as described elsewhere.13 In order to identify whether there were any translocations involving the immunoglobulin (Ig) heavy and light chain loci, FISH assays were conducted with LSI IGH Dual Color Break Apart probe and IGL break apart probe.10 A noncommercial probe was used for the analysis of Xq21-qter duplication (Supplementary Table 3). Total Chromosome DNA X Probe (Qbiogene, Carlsbad, CA, USA) was used in combination with LSI IGH Dual Color Break Apart probe (Vysis), in order to study the involvement of IGH in Xq21 translocation. The probes used for the validation of amplification and homozygous deletions are described in supplementary Table 3.
SKY was carried out as previously described.13,14 At least five metaphases were analyzed for each case.
A high number of genomic aberrations were found in MM
The presence of DNA copy number changes in 26 primary MM samples was analyzed by high resolution array CGH. Most of the samples had a normal karyotype at diagnosis (Table 1); however, genomic copy number analysis, performed on CD138 cells, allowed the identification of CNV in 100% of the cases. A total of 270 copy number changes (ranging from 2 to 26) were found; 145 were copy number gains and 125 were losses. The median size of the aberration was 36 Mb (range, 0.026 Mb to 199 Mb). Gains most frequently involved chromosomes 19 (65%), 9 (42%), 15 (42%), 5 (38%), 3 (38%), 7 (35%) and 1q (38%), and losses in 13q (38%), 8p (35%), 16p (31%), 22q (31%) and 1p (27%). The most frequent aberrations were structural rearrangements involving chromosome segments (41.5% of the changes), followed by whole chromosome gains and losses (31.1%). There are two major genetic categories of MM: H-MM with 48 to 74 chromosomes due to specific chromosome gains and NH-MM without this genotype. Fifty percent of our cases were H-MM, the other fifty percent were NH-MM, initially assessed by array CGH and confirmed by conventional FISH (Table 1). Furthermore, unsupervised clustering, carried out with the most frequent aberrations found in our series, segregated two groups belonging to H-MM and NH-MM (Supplementary Figure 1). When H-MM and NH-MM cases were compared, we found that numerical changes were significantly more frequent in H-MM samples (p<0.001) due to a higher frequency of whole chromosome gains (p<0.001). However, monosomies seemed to be more frequent in the NH-MM group (p=0.076). As far as structural aberrations, we found that gains >3Mb were significantly more frequent (p=0.037) in H-MM cases (Supplementary Table 3). Other types of changes were equally distributed in both groups.
Clustering analysis identified two different H-MM subgroups
In order to obtain a reliable picture of common aberrations present in this series, we defined the SORI for gains and losses. Sixty-eight SORI were identified: 33 gains and 35 losses (Table 2). The most frequent gained SORI was 19p13.12-pter (65%) followed by 9q34, 19q13 and 19q13.2 (50%) and the most frequent losses were in 13q12-q21 (38%), 13q12.12 (35%), 8p12-p22 (27%) and 16p13.12 (23%). The median size was 1.5 Mb ranging from 0.026 Mb (a few genes) to 102 Mb (whole chromosomes). Deletion at 16p13.12, present in 23% of the cases, affected only ERCC4, a modifier gene reported to be involved in breast cancer.15 Unsupervised and supervised clustering analyses were performed using SORI as variables (Figure 1A). Unsupervised analysis was able to segregate H-MM from NH-MM, and after a supervised analysis, gained regions in chromosomes 15, 5, 19, 3, 9, 7 and 11q were found to be significantly (adjusted p-value <0.05) more frequent in the H-MM group. In the unsupervised analysis, our H-MM cases appeared to be distributed in two subgroups that statistically differed (adjusted p-value=0.03, Fisher’s exact test) according to the presence of gains in chromosome 7. Similarly, the unsupervised clustering analysis segregated two different subgroups within the NH-MM cases. One of the clusters, formed of only two cases, was characterized by the presence of a high number of CNV, such as gains in 8q24.21 (C-MYC) and absence of IGH translocations.
New genomic aberrations in MM
High level amplifications and homozygous deletions represented 7% of the observed CNV. They appeared in H-MM and NH-MM, but, in general, clustered in cases that accumulated a high number of genomic aberrations. Nine high level amplifcations were found and most of these regions were also gained (Table 3) in other cases from our series. Among the genes contained in all the amplified regions, 31% are involved in signal transduction (GO:0007165), such as DIRAS2, ADORA2B, TNFRSF13B, RALGDS, CCNE1 and SYK. Six homozygous deletions were observed (Table 3). Reinforcing the putative pathogenic role of this finding, these regions also showed up as single copy losses in several additional samples of our series. Of interest, we observed a 376 Kb homozygous deletions in chromosome 13, where four genes (HSMPP8, PSPC1, ZNF237, ZNF198) are located (Table 2). This deletion was observed in two cases (#15 and 20). Case #20 showed a single copy of chromosome 13 that was involved in a translocation with chromosome 14 (Figure 1). Case #15 showed a homozygous deletions that included two consecutive clones involving only ZNF198. In general, the genes are involved in cellular physiological processes (GO:0050875), although some of them are more specifically involved in cell death (GO:0008219) and signal transduction (GO:0007165), such as BIRC2, BIRC3 and MAP3K5; and others are metalloendopeptidases (GO:0008237) such as MMP8, MMP10, MMP1 and MMP13. These high level amplifications and homozygous deletions were validated by FISH assays using BAC clones covering genes in lost or gained regions (Supplementary Figure 2).
A novel duplication on Xq21.33 was identified in five out of the 26 primary cases and in the MM cell line L363. FISH studies allowed the mapping of the duplication breakpoints that clustered around a 10 Mb region in Xq21.33 (from nearly 86122579bp to 98277428 bp). SKY analysis (Figure 1) demonstrated that duplications were the unbalanced outcome of translocations with breakpoints at Xq21: in L363 involving chromosome 14q, t(X;14)(q21.33; q31.1); and in case #20 involving chromosome 21; t(X;21)(q21.33;q22.3). The breakpoint in chromosome 14 was revealed out by the array CGH profile and we confirmed that IGH was not involved by using a combination of a chromosome X painting probe (Qbiogene) and the LSI IGH Dual Color Break Apart probe (Vysis) in the same hybridization (Supplementary Figure 2).
Multiple myeloma is an incurable malignant plasma cell neoplasia characterized by the accumulation of malignant plasma cells in the bone marrow. Genetically, MM is defined by a large and variable number of genetic and chromosome rearrangements. In general, two major genetic subtypes of MM have been defined: the hyperdiploid variant associated with multiple chromosome trisomies and the non-hyperdiploid variant with a high prevalence of IGH translocations.5 While the biology of MM bearing IGH translocations has been extensively studied,16 little is known about the impact of genomic gains and losses in the pathogenesis of MM. In this regard, candidate genes, overexpressed in amplified chromosomal regions in MM cells lines have recently been proposed.9 These events have been described in other types of cancer and have been used as therapeutic targets.
In this study, an array CGH platform was used for the analysis of copy number imbalances in CD138 plasma cells. Although, as expected, most of the samples showed an unaltered karyotype at diagnosis, we found CNV in all of them. Observed frequencies of CNV, losses of chromosome 13q and duplications of chromosome 1q, and Ig genes translocations, were concordant with those in previous studies.6,7,9,12,17 Array CGH enabled efficient segregation of the series into H-MM and NH-MM cases; half of the cases were H-MM and half NH-MM. The presence of IGH translocations was associated with NH-MM; among the H-MM cases, only one showed IGH translocations and two IGL translocations. It has been suggested that translocations present in H-MM are secondary events.5,17
Array CGH provides a level of resolution that overcomes the limits of chromosome-based CGH which is, in the best conditions, near 3 Mb. Due to the high resolution of the platform (44K) we identified several aberrations smaller than this limit, distributed in both H-MM and NH-MM cases, which had not been previously described. We delimited several SORI, offering a reliable picture of the aberrations occurring in MM.12 Apart from segregating the two major genetic categories of MM, the unsupervised clustering based on the SORI aberration (Figure 1A) identified two H-MM subgroups, one of which was associated with chromosome 7 gains. Gains in 7q have been described in Burkitt′s lymphoma,18 a combination in which they were associated with a poor outcome.
The approach used also allowed the identification of novel genomic markers and the definition of high level amplifications and homozygous deletions. Nine high level amplifications were identified. Some of them, such as 9q22.2, 19q13 and 16q22.2, had not been previously described in MM, but may harbor candidate oncogenes, as we have demonstrated in mantle cell lymphoma, in which we observed primary samples showing amplification and overexpression of SYK,19 or CCNE1, in 19q13, which has been shown to be amplified and overexpressed in endometrial and bladder neoplasia.20,21 Most of the homozygous deletions we found had not been previously described in MM. Reinforcing the putative pathogenic role of this finding, these regions also showed up as single copy losses in several additional samples of our series. As regards homozygous deletion in chromosome 13, ZNF198 seems to be a good candidate for the target of the deletion because, besides the deletion described in case #20, in case #15, the homozygous deletion covered only the ZNF198 gene. Furthermore, candidate genes such as PORIMIN, YAP1 and MMP8, which are located in the homozygous deletion that we found at 11q22.2, were reported to be deleted and down regulated in a previous study. Five cases showed duplication on Xq21.33. Although some CGH studies have reported data on Xq aberrations,22,23 we provide here the first description of the frequency and genetic origin of this marker in MM. In summary, our study represents the analysis of CNV in MM at the highest level of resolution ever reported. Our work includes some data confirming results obtained with other technological approaches, and provides a complete genomic characterization of the two genetic categories of MM. Additionally, we have demonstrated for the first time that extreme genomic aberrations, such as high level amplifications and homozygous deletions, are frequent events in MM. We believe that these data warrant further investigation to gain a deeper insight into the role of genomic instability in MM and its potential use for therapeutic purposes.
we thank Drs Marta R. Salazar (University Clínic, Pamplona), Maite Ardanaz (Hospital Txagorritxu, Vitoria), Jose Ferreiro (Hospital Donostia, San Sebastian), Filomena Floristán (Hospital de Cruces, Bilbao), Jose A Marquez (Hospital de Basurto, Bilbao), Carmen Mateos (Hospital Virgen del Camino, Pamplona), Ana Gorosquieta (Hospital de Navarra, Pamplona) and Gotzon Pereda (Hospital de Santiago, Vitoria) for kindly providing samples and clinical data. We are also grateful to M Carmen Martín Guijarro, Carmen Carralero and Gloria Soler for their excellent technical support
- Authors’ Contributions CL carried out all technical arrays, SKY and some FISH experiments. She contributed to the writing of the paper; BS performed FISH assays and contributed to the writing the paper; SA designed the study and contributed to the microarray analisis of the data; JS, BF and DB carried out microarray analyses and contributed to the analysis of the data; FP has sustantially contributed substantially with biological material and clinical data; MJC contributed with essential biological material, cytogenetic information and helped to design the experiments; JCC designed the work, collected all data and wrote the paper.
- Conflict of Interest The authors reported no potential conflicts of interest.
- Funding: CL and BS have PhD fellowships from Gobierno de Navarra. JS has a PhD fellowship from Ministerio de Educación y Ciencia. BF has a Marie Curie PhD Early Stage Research Training Fellowship. Work funded by Red Temática FIS G03/136 Mieloma Múltiple y otras Ganmapatías: de la Génesis a la Terapéutica and by Grant PI040555 from Fondo Investigaciones Sanitarias, Ministerio de Sanidad y Consumo to JCC.
- Received November 16, 2006.
- Accepted March 15, 2007.
- Kyle RA, Rajkumar SV. Multiple myeloma. N Engl J Med. 2004; 351:1860-73. Google Scholar
- Seidl S, Kaufmann H, Drach J. New insights into the pathophysiology of multiple myeloma. Lancet Oncol. 2003; 4:557-64. Google Scholar
- Kuehl WM, Bergsagel PL. Multiple myeloma: evolving genetic events and host interactions. Nat Rev Cancer. 2002; 2:175-87. Google Scholar
- Fonseca R, Barlogie B, Bataille R, Bastard C, Bergsagel PL, Chesi M. Genetics and cytogenetics of multiple myeloma: a workshop report. Cancer Res. 2004; 64:1546-58. Google Scholar
- Chng WJ, Santana-Davila R, Van Wier SA, Ahmann GJ, Jalal SM, Bergsagel PL. Prognostic factors for hyperdiploid-myeloma: effects of chromosome 13 deletions and IgH translocations. Leukemia. 2006; 20:807-13. Google Scholar
- Cigudosa JC, Rao PH, Calasanz MJ, Odero MD, Michaeli J, Jhanwar SC. Characterization of nonrandom chromosomal gains and losses in multiple myeloma by comparative genomic hybridization. Blood. 1998; 91:3007-10. Google Scholar
- Gutierrez NC, Garcia JL, Hernandez JM, Lumbreras E, Castellanos M, Rasillo A. Prognostic and biologic significance of chromosomal imbalances assessed by comparative genomic hybridization in multiple myeloma. Blood. 2004; 104:2661-6. Google Scholar
- Rao PH. Comparative genomic hybridization for analysis of changes in DNA copy number in multiple myeloma. Methods Mol Med. 2005; 113:71-83. Google Scholar
- Largo C, Alvarez S, Saez B, Blesa D, Martin-Subero JI, Gonzalez-Garcia I. Identification of overexpressed genes in frequently gained/amplified chromosome regions in multiple myeloma. Haematologica. 2006; 91:184-91. Google Scholar
- Martin-Subero JI, Harder L, Gesk S, Schlegelberger B, Grote W, Martinez-Climent JA. Interphase FISH assays for the detection of translocations with breakpoints in immunoglobulin light chain loci. Int J Cancer. 2002; 98:470-4. Google Scholar
- Barrett MT, Scheffer A, Ben-Dor A, Sampas N, Lipson D, Kincaid R. Comparative genomic hybridization using oligonucleotide microarrays and total genomic DNA. Proc Natl Acad Sci USA. 2004; 101:17765-70. Google Scholar
- Carrasco DR, Tonon G, Huang Y, Zhang Y, Sinha R, Feng B. High-resolution genomic profiles define distinct clinico-pathogenetic subgroups of multiple myeloma patients. Cancer Cell. 2006; 9:313-25. Google Scholar
- Rodriguez-Perales S, Martinez-Ramirez A, de Andres SA, Valle L, Urioste M, Benitez J. Molecular cytogenetic characterization of rhabdomyosarcoma cell lines. Cancer Genet Cytogenet. 2004; 148:35-43. Google Scholar
- Rao PH, Cigudosa JC, Ning Y, Calasanz MJ, Iida S, Tagawa S. Multicolor spectral karyotyping identifies new recurring breakpoints and translocations in multiple myeloma. Blood. 1998; 92:1743-8. Google Scholar
- Smith TR, Levine EA, Perrier ND, Miller MS, Freimanis RI, Lohman K. DNA-repair genetic polymorphisms and breast cancer risk. Cancer Epidemiol Biomarkers Prev. 2003; 12:1200-4. Google Scholar
- Bergsagel PL, Kuehl WM. Molecular pathogenesis and a consequent classification of multiple myeloma. J Clin Oncol. 2005; 23:6333-8. Google Scholar
- Smadja NV, Fruchart C, Isnard F, Louvet C, Dutel JL, Cheron N. Chromosomal analysis in multiple myeloma: cytogenetic evidence of two different diseases. Leukemia. 1998; 12:960-9. Google Scholar
- Garcia JL, Hernandez JM, Gutierrez NC, Flores T, Gonzalez D, Calasanz MJ. Abnormalities on 1q and 7q are associated with poor outcome in sporadic Burkitt's lymphoma. A cytogenetic and comparative genomic hybridization study. Leukemia. 2003; 17:2016-24. Google Scholar
- Rinaldi A, Kwee I, Taborelli M, Largo C, Uccella S, Martin V. Genomic and expression profiling identifies the B-cell associated tyrosine kinase Syk as a possible therapeutic target in mantle cell lymphoma. Br J Haematol. 2006; 132:303-16. Google Scholar
- Cassia R, Moreno-Bueno G, Rodriguez-Perales S, Hardisson D, Cigudosa JC, Palacios J. Cyclin E gene (CCNE) amplification and hCDC4 mutations in endometrial carcinoma. J Pathol. 2003; 201:589-95. Google Scholar
- Richter J, Wagner U, Kononen J, Fijan A, Bruderer J, Schmid U. High-throughput tissue microarray analysis of cyclin E gene amplification and overexpression in urinary bladder cancer. Am J Pathol. 2000; 157:787-94. Google Scholar
- Inoue J, Otsuki T, Hirasawa A, Imoto I, Matsuo Y, Shimizu S. Overexpression of PDZK1 within the 1q12-q22 amplicon is likely to be associated with drug-resistance phenotype in multiple myeloma. Am J Pathol. 2004; 165:71-81. Google Scholar
- Liebisch P, Viardot A, Bassermann N, Wendl C, Roth K, Goldschmidt H. Value of comparative genomic hybridization and fluorescence in situ hybridization for molecular diagnostics in multiple myeloma. Br J Haematol. 2003; 122:193-201. Google Scholar