Integrative Genomics Analyses Reveal Molecularly Distinct ... · Integrative Genomics Analyses Reveal Molecularly Distinct Subgroups of B-Cell Chronic Lymphocytic Leukemia Patients

Human Cancer Biology

Integrative Genomics Analyses Reveal Molecularly Distinct Subgroupsof B-Cell Chronic Lymphocytic Leukemia Patients with 13q14 Deletion

Laura Mosca1, Sonia Fabris1, Marta Lionetti1, Katia Todoerti1, Luca Agnelli1, Fortunato Morabito2,Giovanna Cutrona3, Adrian Andronache1, Serena Matis3, Francesco Ferrari4, Massimo Gentile2,Mauro Spriano5, Vincenzo Callea6, Gianluca Festini7, Stefano Molica8, Giorgio Lambertenghi Deliliers1,Silvio Bicciato4, Manlio Ferrarini3,9, and Antonino Neri1

AbstractPurpose: Chromosome 13q14 deletion occurs in a substantial number of chronic lymphocytic

leukemia (CLL) patients and it is believed to play a pathogenetic role. The exact mechanisms involved

in this lesion have not yet been fully elucidated because of its heterogeneity and the imprecise knowledge of

the implicated genes. This study was addressed to further contribute to the molecular definition of this

lesion in CLL.

Experimental Design: We applied single-nucleotide polymorphism (SNP)-array technology and gene

expression profiling data to investigate the 13q14 deletion occurring in a panel of 100 untreated, early-stage

(Binet A) patients representative of the major genetics, molecular, and biological features of the disease.

Results: Concordantly with FISH analysis, SNP arrays identified 44 patients with del(13)(q14)

including 11 cases with a biallelic deletion. The shorter monoallelic deletion was 635-kb long. The loss

of the miR-15a/16-1 cluster occurred in all del(13)(q14) cases except in 2 patients with a monoallelic

deletion, who retained both copies. MiR-15a/16 expression was significantly downregulated only in

patients with the biallelic loss of the miRNA cluster compared to 13q normal cases. Finally, the natural

grouping of SNP profiles by nonnegative matrix factorization algorithm showed that patients could be

classified into 2 separate clusters, mainly characterized by short/biallelic versus wide/monoallelic 13q14

deletions. Supervised analyses of expression data showed that specific transcriptional profiles are correlated

with these 2 genomic subgroups.

Conclusions: Overall, our data highlight the presence of 2 distinct molecular types of 13q14 deletions,

which may be of clinical relevance in CLL. Clin Cancer Res; 16(23); 5641–53. �2010 AACR.

B-cell chronic lymphocytic leukemia (CLL) is a lympho-proliferative disorder characterized by a variable clinicalcourse: some patients progress rapidly toward more

advanced stages whereas others survive for a long timewithout disease progression (1). Recently, considerableefforts have been addressed to the identification of geno-mic aberrations that could explain the pathogeneticmechanisms and the clinical heterogeneity of the disease(2–5). Deletions of 13q14, 11q22.3, and 17p13, andtrisomy 12 are common in CLL and may play a role inpathogenesis and disease progression (4). 13q14 deletionoccurs in approximately 50% of CLL and is associated witha favorable prognosis when present as the sole abnormality(6). Notably, 2 microRNA genes, miR-15a and miR-16-1,located at 13q14, have been reported to be downregulatedin del(13)(q14) patients (7) and strongly suggest theirpotential role in the disease. However, the 13q14 deletionis not always accompanied by defects in miR-15a/16-1expression, suggesting a more complex heterogeneity ofthe deletion itself (8, 9).

The recent introduction of microarray technology hasimproved the possibility of combining genome-wideDNA analyses with transcriptomic profiles, thus allowingthe identificationof potential candidate tumorgenes relatedto underlying chromosomal alterations. To furtherelucidate the genomic complexity of the 13q14 deletion

Authors' Affiliations: 1Dipartimento di Scienze Mediche, Universit�a diMilano, U.O. Ematologia 1, Fondazione IRCCS Ca’ Granda OspedaleMaggiore Policlinico, Milan, Italy; 2U.O.C. di Ematologia, Azienda Ospe-daliera di Cosenza, Italy; 3Divisione di Oncologia Medica C, IstitutoNazionale per la Ricerca sul Cancro, IST, Genoa, Italy; 4Dipartimento diScienze Biomediche, Universit�a di Modena e Reggio Emilia, Modena, Italy;5Dipartimento di Ematologia, Azienda Ospedaliera S. Martino, Genoa,Italy; 6Divisione di Ematologia, Azienda Ospedaliera, Reggio Calabria,Italy; 7Centro di Riferimento Ematologico-Seconda Medicina-AziendaOspedaliero-Universitaria, Ospedali Riuniti, Trieste, Italy; 8U.O.C. diOncologia, Azienda Ospedaliera "Pugliese-Ciaccio", Catanzaro, Italy;and 9Dipartimento di Oncologia, Biologia e Genetica, Universit�a degliStudi di Genova, Genoa, Italy

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

L. Mosca and S. Fabris contributed equally to this work.

Corresponding Author: Prof. Antonino Neri, Department of MedicalSciences, University of Milan, Via F. Sforza 35, 20122 Milano, Italy. Phone.þ39.02.55033328; Fax: þ39.02.50320403. E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-10-0151

�2010 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 5641

Research. on October 18, 2020. © 2010 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

Published OnlineFirst October 14, 2010; DOI: 10.1158/1078-0432.CCR-10-0151

http://clincancerres.aacrjournals.org/

in CLL, we applied single-nucleotide polymorphism (SNP)-array technology to a panel of 100 untreated patients withBinet stageAdisease including44patientswith13qdeletionas assessed by fluorescence in situ hybridization (FISH). Wethen investigated the expression levels of the miR-15a/16cluster by quantitative real-time PCR (qRT-PCR) and globalgene expression profiling in a representative panel of cases(see flow chart in Supplementary Figure S1). The integrationbetween genomic and expression data allowed the identi-fication of 2 distinct molecular groups of patients with13q14 deletions. Our data may have implications for thebiology and the prognostic stratification of CLL.

Materials and Methods

PatientsThe study included samples from 100 untreated CLL

patients in Binet stage A: 60 from a retrospective collabora-tive Italian study (10) and 40 enrolled in an Italian pro-spective multicenter study (GISL O-CLL1). Eligibilityrequired a diagnosis of a typical CLL phenotype: CD5/CD19þ and CD23þ, weak surface immunoglobulin (sIg),

and the monotypical expression of k or l light chains byneoplastic cells. Patients were selected to provide a goodnumerical representation of cytogenetic lesions, as assessedby FISH, and IgVH mutation status. All the patients gavetheir informed consent in accordance with our institutionalguidelines.

Sample preparation, immunophenotype, andprognostic markers

Peripheral blood mononuclear cells were isolated byFicoll–Hypaque gradient centrifugation (Seromed), and

CD38 and ZAP-70 expression were determined by flowcytometry (11). IgVH gene usage and mutational statuswere established as previously described (12), with a 2%cutoff value being used to define mutated and unmutatedpatients. For the microarray analyses, the CLL cells wereenriched by negative selection when less than 90% (11).

Fluorescence in situ hybridizationThe most common genomic aberrations, del(17)(p13),

del(11)(q23), del(6)(q23), del(13)(q14), and trisomy 12,were investigated by interphase FISH hybridization. All ofthe probes are commercially available (Vysis; ref. 13).

High-density SNP arrays and data analysisTotal genomic DNA (250 ng) were processed in accor-

dance with the manufacturer’s protocol (Affymetrix),hybridized using Affymetrix GeneChip Human Mapping250K NspI microarrays, and subsequently scanned using aGeneChip Scanner 3000 7G. The images were acquiredusing the Affymetrix GeneChip Operating System (GCOS1.4). The accuracy of the SNP array data were confirmed bythemean andmedian call rates of 95.75% and 96.12% (thequality control specification for 250K arrays is a call rategreater than 93%), respectively.

The entire procedure for the copy number (CN) estima-tion has been fully described previously (14). Briefly, theraw data relating to the individual SNPs were extractedfrom CEL files and converted into signal intensities usingGTYPE 4.1 and Affymetrix Copy Number Analysis Tool(CNAT 4.0.1) software. Each sample was compared with aset of 48 normal Caucasian HapMap references availableon the Affymetrix web site (http://www.affymetrix.com/support/technical/sample_data/500k_data.affx) and thegenomic smoothing window of the Hidden Markov Modelalgorithm was set at 0. After preprocessing, piecewise con-stant estimates of the underlying local DNA CN alterationswere calculated using the DNAcopy Bioconductor package,which looks for optimal breakpoints on the basis of circularbinary segmentation (15), and the median values of theestimated profiles were scaled back to a nominal multi-plicity of 2. After scaling, we determined all the clustersappearing on the frequency distribution (histogram) of theSNP values for all the SNP probes using the k-meansalgorithm on the cumulative profile of all of the arraydata. The transition from one cluster to the other, namelythe threshold from one CN to another, was then estimatedas the meeting point of 2 Gaussian curves and approximat-ing the distributions of mean m and variance s2 of the 2neighboring clusters of SNPs identified in the final cluster-ing. Thus, the thresholds result as follows: inferred CNs ofmore than 2.1 and 2.5 corresponded to gain and amplifi-cation, whereas CNs of less than 1.9 and 1.34 correspondedto loss and biallelic deletion. After segmentation, the SNPdataset was compressed to 1346 probes by eliminatingredundant probes to better balance less represented regionswith those showing a large number of probes, as previouslydescribed (14). A probe was defined as redundant whenit showed CN values identical to those of the most

Translational Relevance

The 13q14 deletion represents the most commongenomic aberration in CLL (50%). Although it is asso-ciated with a favorable prognosis when present as thesole abnormality, the imprecise knowledge of the genesimplicated and its genetic heterogeneity have limitedour understanding on the pathogenetic mechanismscontributing to the disease. Microarray technologyhas improved the possibility of combining genome-wide DNA with transcriptomic profiles to identifypotential candidate tumor genes related to underlyingchromosomal aberrations. Based on SNP array, ourstudy shows the presence of 2 distinct molecular groupsof patients with del(13)(q14) based on the deletion sizeand the presence of biallelic deletions. Notably, globalgene expression profiling identified a significant tran-scriptional deregulation specifically associated with the2 groups. Our data highlight the presence of 2 distinctmolecular types of 13q14 deletion that may be ofclinical relevance for the biology and the prognosticstratification of the disease.

Mosca et al.

Clin Cancer Res; 16(23) December 1, 2010 Clinical Cancer Research5642




contiguous upstream probe in all samples. Then a non-negative matrix factorization (NMF) procedure adapted toR from the original MATLAB package (16) was used toevaluate the meaningful clusters across the whole dataset.For each factor level from 2 to10, NMF was repeated 100times to build a consensus matrix, and the samples wereassigned to clusters on the basis of the consensus results.

Sequence copy number determination by quantitativereal-time PCRReal-time PCR was performed according to a published

protocol (17) using the ABI Prism 7900 sequence detec-tion system. Singleplex amplification reactions were car-ried out in triplicate using 40 ng of template DNA, 1XTaqMan Universal Master Mix, no AmpErase UNG, and a1X Primer-Probe Mix (Applied Biosystems) containingsequence-specific primers and a fluorogenic probe.The TaqMan RNase P Detection Reagents kit and aCustom TaqMan Gene Expression Assay (forward primer:50-GCAATGTCAGCAGTGCCTTAG-30; reverse primer:50-CAGCAGCACAGTTAATACTGGAGAT-30; probe: 50-FAM-CAGCACGTAAATATTG-30) were used to amplifythe ribonuclease P RNA component H1 gene mappedwithin 14q11.2 (present in 2 copies in all of the subjectsand thus used as an internal standard) and the miR-15a/16-1 cluster. The comparative DCt method was used forquantification purposes. DNA samples from 10 controlindividuals with 2 copies of the commonly deleted regionwere selected for PCR calibration. The estimated haploidgene CN was given by the formula 2�DDCt, and the pre-dicted CN was calculated as the closest integer number tothe estimated CN (determining Gene Copy Number usingTaqMan Real-Time PCR Assays on the 7900 HT-QuickReference Card; Applied Biosystems; ref. 18).

Quantification of specific gene expressionby Q-RT-PCRQ-RT-PCR of specific genes (TPT1, WBP4, PEA15,

and LGALS1) and mature miRNAs (hsa-miR-15a andhsa-miR-16) was performed using commercial TaqManassays (Applied Biosystems) as previously described(19). The relative gene and miRNA expression levels werecalculated using the 2�DCt method (Applied BiosystemsUser Bullettin No. 2) as previously described (19).Pearson’s correlation coefficient was calculated to testcorrelation between gene expression and Q-RT-PCR data.

Gene expression profilingTwenty-two CLL samples with 13q deletion included in

the retrospective database underwent gene expressionprofiling (GEP) analysis. Total RNA was extracted usingthe TRIZOL reagent (Invitrogen) and purified using theRNeasy total RNA Isolation Kit (Qiagen). The biotin-labeled complementary RNA was prepared and hybri-dized with GeneChip Human Genome U133A Arrays(Affymetrix Inc.), which were scanned (GeneChip Scan-ner 3000 7G; Affymetrix Inc.) in accordance with themanufacturer’s protocols. The probe data were converted

to expression values using the Bioconductor function forthe robust multiarray average procedure, as describedpreviously (20). Supervised analysis was made usingSignificant Analysis of Microarrays software, version3.0.2 (http://www-stat.stanford.edu/�tibs/SAM/index.html/; ref. 21). The cutoff point for significance at amedian false discovery rate (FDR) less than 5% (i.e., q< 0.05) was determined by tuning the D parameter on theFDR and controlling the q-value of the selected probes.dChip software (22) was used to represent the selectedprobe lists; NetAffx (https://www.affymetrix.com/analy-sis/netaffx/) was used for the functional annotation studyof the list. The genotyping and gene expression data areavailable at NCBI’s Gene Expression Omnibus throughGEO Series Accession no. GSE16746.

Statistical analysisThe data were statistically analyzed using conventional

procedures in R software (Kruskal–Wallis tests, Kendall’s tcorrelations, Fisher’s exact tests, and q-value calculations).

Results

Molecular characteristics of the CLL patientsThemain characteristics of the 100CLL patients included

in the study are reported in Supplementary Tables S1 andS2. Based on FISH analyses, 44 cases carried the 13q14deletion: 34 as the sole abnormality, and the remainingshowed 11q22.3 (6 patients), 17p13.1 (2 patients),11q22.3 and 17p13.1 deletions (1 patient), or 6q23 dele-tion (1 patient). Biallelic 13q14 deletions were identified in11 cases, 5 of which (CS3, CZ42, GE110, CS0100 andLD0062) showed the presence of subclones characterizedby the biallelic deletion (ranging from 68% to 83.5%).Trisomy 12 was present in 21 patients as a sole abnorm-ality. Overall, the 11q22.3, 17p13.1, and 6q23 deletionswere present in 15, 7, and 2 patients, respectively. Fifty-fivecases had unmutated IgVH genes; ZAP-70 and CD38 wereexpressed in 42 and 46 cases, respectively. 13q deletion waspresent in 25 of 45 (55.5%) patients with mutated IgVHgenes, 15 of 42 (35.7%) of the ZAP-70 positive patients,and 15 of 46 (32.6%) of the CD38 positive patients.

SNP array data were concordant with FISH results(Supplementary Tables S2 and S3). In addition, SNP arraysdetected short deletions involving the second 13q allele in3 del(13)(q14) patients (TS12, VB0013, and PS0044);deletions were, respectively, 467, 468, and 291-kb longand located centromerically to the LSI D13S25 FISH probe(Fig. 1A and B). Furthermore, SNP arrays revealed a 6qdeletion of approximately 39 Mb in length and locatedupstream of the LSI MYB FISH probe in a single patients(CZ47; data not shown). These aberrations were confirmedby specific FISH probes in all cases (data not shown).

Characterization of the 13q deletion by SNP arraysThe SNP arrays showed that the 13q deletions varied

considerably in size, ranging from a minimum of 291 kb(PS0044) to a maximum of 56 Mb (CZ36; Fig. 1A and B).

Integrative Genomics Analysis of 13q14-Deleted CLL

www.aacrjournals.org Clin Cancer Res; 16(23) December 1, 2010 5643




C

A

B

Fig. 1. Chromosome 13 deletion pattern in 44 CLLs. A, monoallelic (gray lines) and biallelic losses (black lines) in the 44 deleted patients. B, theenlarged subregion spanning 13q14.2–q14.3 between physical positions 44.50 and 52.00 Mb, including the minimal monoallelic deletion. C, thelocalization of SNPs and the FISH probe encompassing the minimally altered region (gray bar). Gene locations and transcriptional orientation areindicated by the horizontal arrows at the bottom.

Mosca et al.





The minimal monoallelic deletion was 635-kb longspanning from SNP_A-2003314 to SNP_A-2003318(physical position 49,635,024 bp to 50,270,550 bp;Fig. 1C); notably, the miR-15a/16-1 cluster is locatedupstream (�87 kb) of the centromeric SNP_A-2003314. With regard to the miR-15a/16-1 cluster, SNParray analysis depicted the following scenario: (i) reten-tion of 2 copies of the cluster in 56 patients with a normal

13q and in 2 cases (CD0018 and CS95) showing a 13qmonoallelic deletion telomeric to the cluster; (ii) reten-tion of 1 copy of the cluster in 29 patients with a 13qmonoallelic deletion and in 2 cases (CZ42 and PS0044)showing a deletion telomeric to the cluster in the secondallele; (iii) loss of both copies of the cluster in theremaining 11 patients showing biallelic deletions(Fig. 1A and B; Supplementary Table S4).

C

A

B

Fig. 2. Sequence copy number and expression quantification of miR-15a/16 as assessed by Q-RT-PCR. A, vertical axis: 2�DDCt values for the 32 investigatedpatients. The inferred CN values obtained on the basis of the criteria described in Material and Methods are shown for each patientsas white bars (CN ¼ 0), striped bars (CN ¼ 1), and black bars (CN ¼ 2); horizontal axis: patient distribution into 3 classes on the basis of the numberof alleles (0, 1, and 2) detected by SNP array. B and C, expression of miR-15a and miR-16 as calculated using the 2�DCt method and ranked accordingto the corresponding inferred CN value (CN0 ¼ 11 cases; CN1 ¼ 28 cases, including PS0044 patient; CN2 ¼ 39 cases, including CD0018 and CS95patients). P values of the comparison of between the biallelic and monoallelic groups.






The status of the miR-15a/16-1 cluster was investigatedfurther using a custom Q-RT-PCR assay on DNA from 32patients: 10 with biallelic deletions, 10 with monoallelicdeletions, and 12 normal at 13q14 based on SNP arraydata. As shown in Figure 2A and B and SupplementaryTable S4, the estimated CN obtained with this approachwere concordant with the SNP array data (P < 0.0001)with 3 exceptions: 2 patients (CS3 and GE110) classifiedas CN ¼ 1 by RT-PCR and CN ¼ 0 by SNP array, whoshowed 2 distinct cell populations by FISH characterizedby either mono- or biallelic losses, and 1 patient (RC21)classified as normal by RT-PCR and CN ¼ 1 by SNP arraywho showed a monoallelic loss in a fraction of interphasenuclei (61%; Fig. 2A). These findings suggest that clonalheterogeneity may account for the discrepancies betweenthe real-time PCR and SNP array data.

Furthermore, we compared miR-15a and miR-16expressions in 78 of 100 cases including 11 with biallelic

deletions, 28 with monoallelic deletions and 39 normalcases based on SNP array CN values. Cases with biallelicdeletions had a miRNA expression significantly lowerthan cases retaining 1 or 2 copies (P < 0.0001 for miR-15a, P ¼ 0.0001 for miR-16) whereas no statisticallysignificant differences in miRNA expression levels couldbe observed between normal cases and those retaining asingle copy of the cluster (Fig. 2B and C; SupplementaryTable S5).

Finally, SNP array documented a loss of the RB1 gene in28 of 44 cases with 13q14 deletion (63.6%). Two samples(CS0100 and TS12) showed a biallelic loss of the RB1locus, which was associated with a homozygous deletion ofthe 13q14 FISH probe (case CS0100) or was the result of asmall deletion involving the RB1 locus on the second allele(case TS12; Fig. 1B). All these findings were validated byFISH using the RB1 specific clone RP11-305D15 (data notshown).

Fig. 3. Heat map of significantly altered DNA regions in 100 CLLs, as assessed by GeneChip Human Mapping 250K Nsp. The genomic profiles of the CLLsamples (horizontal lines) are clustered into 4 groups in accordance with the NMF method. Horizontal axis: chromosome localization. The dashed linesrepresent the centromeres. Light-green, biallelic deletion; dark-green, loss; white, normal copy number; red, gain.

Mosca et al.





Identification of 2 genetically distinct patientsubgroups with del(13)(q14)To identify the most significant natural grouping of

genome profiles in our panel, we used the NMF algorithm.This analysis led to the identification of 4 major groups(correlation coefficient ¼ 0.95) characterized by 13q dele-tion (groups I and II, 27 and 13 cases, respectively) and bytrisomy 12 (group III, 21 cases), whereas no specific altera-tion could be associated with group IV (39 cases; Fig. 3;Table 1). Patients from groups I and II differed in thedeletion size and for the occurrence of mono- or biallelicdeletions: group I included samples with relatively smallerlosses, 10 of which showed biallelic deletions, whereasgroup II included samples with larger losses, all but one(GE110) showing monoallelic deletions. The presence of 4samples with 13q14 deletions in group IV was in all like-lihood related to the fact that the inferred CN values (ran-ging from 1.74 to 1.85) were very close to the threshold(1.9) between monoallelic deletion and the retention of 2copies. These "marginal" CNdata are consistent with clonalheterogeneity and are in accordance with the small percen-tage of interphase nuclei detected by FISH showing 13q14deletion (no more than 21% in all cases), which may affectthe clustering analysis. RB1 deletion occurred in 14 of 27cases in group I (52%), 12of 13 in group II (92%), and2of 4in group IV. Only 10 of the 44 patients with del(13)(q14)carried additional known genetic lesions, namely 11q22.3(6 patients), 17p13.1 (2 patients), 11q22.3 plus 17p13.1(1 patient), 6q23 deletions (1 patient); however, no corre-lation was found with the 2 specific subgroups I and II.Finally, there was no significant association between caseswithinNMF groups I and II andCD38or ZAP-70 expressionor IgVH mutational status (data not shown).

Gene expression patterns associated with distinctgroups of patients with 13q deletionFifteen cases from group I and 7 from group II, for whom

RNA material was available, were profiled on GeneChipHG-U133A arrays. To verify whether the 2 groups could bedivided on the basis of the differences of their expressionprofiles, we made a conventional unsupervised analysisusing hierarchical agglomerative clustering at differentlevels on (i) all of the probes in the array; (ii) all of theprobes mapped to the chromosome 13; or (iii) all ofthe probes within the 13q14 region. Notably, a significantgrouping (i.e., group I and group II patients in 2 separatebranches) was only detected when the 89 probes (65 genes)mapped on the 13q14 region were used (P < 0.0001;

Fig. 4A). This finding indicates that, albeit differences existbetween the 2 groups, these are not sufficient to drive theclustering when the global transcriptional profiling wasconsidered (i.e., the whole matrix). Thus, we carried out asupervised analysis to characterize the specific transcrip-tional profiles distinguishing NMF groups I and II. Weidentified 76 differentially expressed probe sets, specific to63 well-characterized genes (Fig. 4B), 10 of which weredownregulated and 53 upregulated in group II comparedwith group I. Six of the 10 downregulated genes mapped to13q13-q14 and exhibited amonoallelic deletion inmost orall cases within group II [3 of 7 (43%) for FOXO1; 4 of 7(57%) for EXOSC1 and WBP4; 6 of 7 (86%) for TPT1 andNUFIP1; 7 of 7 (100%) for ESD) compared with cases ingroup I (1 of 15 (6.7%) for EXOSC1, TPT1, ESD, FOXO1,and NUFIP1; 2 of 15 (13.3%) forWBP4]. The whole list ofdifferentially expressed genes is reported in Table 2.

We selected 2 downregulated (TPT1 and WBP4) and 2upregulated (PEA15 and LGALS1) genes in group II versusgroup I for Q-RT-PCR validation of the microarray data.The Q-RT-PCR analyses were made in a subset of 20 of 22patients for whom RNA material was available (14 belong-ing to group I, 6 to group II). The correspondence betweenthe microarray and Q-RT-PCR data were evaluated byassessing the correlation coefficients of the expressionlevels determined by the 2 analyses: the coefficients were0.71 for TPT1, 0.79 forWBP4, 0.60 for PEA15, and 0.78 forLGALS1 probe, thus indicating a very good concordance forall of the tested genes (Supplementary Fig. S2).

Discussion

This study focused on the molecular characterization ofthe 13q14 deletion based on SNP arrays and gene expres-sion profiling analyses. Several aspects distinguish thisfrom previous reports (9, 23) including a homogeneouscohort of untreated early-stage patients (Binet A), the use ofhighly purified CLL samples and a higher-resolution SNParray (250K NspI), the application of stringent statistics todefine the genetic groups, and the integration of genomicand gene expression data in a significant number of cases.The main findings relate to the description of the topo-graphy of 13q14 deletions, the reassessment of the fate ofmiR-15a/16-1 cluster in relation to the genomic losses, and,perhaps more importantly, the definition of 2 major geno-mic groups of patients with 13q14 deletions also charac-terized by distinct transcriptional patterns.

Table 1. Molecular characteristics of the 4 genomic groups

NMF cluster No. ofcases

del(11q22.3) Monoallelicdel(13q14)

Biallelicdel(13q14)

del(17p13.1) 12 CD38þ ZAP70þ UnmutatedIgVH

I 27 3 (11.1%) 17 (63.0%) 10 (37.0%) 1 (3.7%) 0 (0%) 8 (29.6%) 10 (37.0%) 11 (40.7%)II 13 4 (30.8%) 12 (92.3%) 1 (7.7%) 2 (15.4%) 0 (0%) 4 (30.8%) 3 (23.1%) 6 (46.1%)III 21 0 (0%) 0 (0%) 0 (0%) 0 (0%) 21 (100%) 17 (81.0%) 12 (57.1%) 15 (71.4%)IV 39 8 (20.5%) 4 (10.3%) 0 (0%) 4 (10.3%) 0 (0%) 17 (43.6%) 17 (43.6%) 23 (59.0%)






Tab

le2.

Func

tiona

lann

otations

ofthe63

gene

siden

tifiedas

differen

tially

expressed

amon

gthe2NMFgrou

psbySAM

2-clas

san

alysis

Probese

tID

Gen

esy

mbol

Gen

etitle

Sco

re(d)

Fold

chan

ge

q(%

)Cytoban

d

2128

69_x

_at

TPT1

Tumor

protein,

tran

slationa

llyco

ntrolled1

�6.330

80.79

013

q12

–q1

4

2090

09_a

tESD

Esteras

eD/formylglutathion

ehy

drolase

�5.013

00.56

013

q14

.1–q14

.2

2151

36_s

_at

EXOSC8

Exo

someco

mpon

ent8

�4.588

80.61

2.19

13q13

.122

2182

_s_a

tCNOT2

CCR4-NOTtran

scrip

tion

complex,

subun

it2

�4.491

10.69

2.19

12q15

2027

23_s

_at

FOXO1

Forkhe

adbox

O1

�4.390

60.67

3.26

13q14

.120

5134

_s_a

tNUFIP1

Nuc

lear

frag

ileXmen

tal

retardationprotein

interactingprotein

1

�4.325

70.66

3.26

13q14

2103

46_s

_at

CLK

4CDC-likekina

se4

�4.314

60.65

3.26

5q35

2132

56_a

tMARCH3

Mem

brane

-ass

ociated

ringfin

ger(C3H

C4)

3�4

.169

60.51

4.20

5q23

.2

2035

97_s

_at

WBP4

WW

dom

ainbinding

protein

4(fo

rmin

binding

protein21

)

�4.125

80.64

4.20

13q14

.11

2041

55_s

_at

QSK

Serine/threon

ine-protein

kina

seQSK

�4.066

20.72

4.72

11q23

.3

2130

11_s

_at

TPI1

Triose

pho

spha

teisom

eras

e1

3.93

551.60

4.72

12p13

2010

13_s

_at

PAICS

Pho

spho

ribos

ylam

inoimidaz

ole

carbox

ylas

e,pho

spho

ribos

ylam

inoimidaz

ole

succ

inoc

arbox

amidesynthe

tase

3.94

071.73

4.72

4q12

2040

47_s

_at

PHACTR

2Pho

spha

tase

andac

tinregu

lator2

3.94

281.33

4.72

6q24

.2

2018

50_a

tCAPG

Cap

pingprotein(actin

filam

ent),

gelsolin-like

3.94

392.24

4.72

2p11

.2

2015

77_a

tNME1

Non

metas

tatic

cells

1,protein

(NM23

A)ex

pres

sedin

3.96

021.62

4.72

17q21

.3

2178

65_a

tRNF1

30Ringfin

gerprotein13

03.96

172.02

4.72

5q35

.320

3857

_s_a

tPDIA5

Protein

disulfid

eisom

eras

efamily

A,mem

ber

53.99

101.25

4.01

3q21

.1

2011

05_a

tLG

ALS

1Le

ctin,ga

lactos

ide-binding,

soluble,1

3.99

673.27

4.01

22q13

.1

4048

9_at

ATN

1Atrop

hin1

4.03

881.23

4.01

12p13

.31

(Con

tinue

don

thefollo

wingpag

e)

Mosca et al.





Tab

le2.

Func

tiona

lan

notatio

nsof

the

63ge

nes

iden

tified

asdifferen

tially

expressed

amon

gthe

2NMF

grou

ps

by

SAM

2-clas

san

alysis

(Con

t'd)

Probese

tID

Gen

esy

mbol

Gen

etitle

Sco

re(d)

Fold

chan

ge

q(%

)Cytoban

d

4412

0_at

ADCK2

aarF

domainco

ntaining

kina

se2

4.06

161.18

4.01

7q32

–q34

2093

29_x

_at

HIG

D2A

HIG

1hy

pox

iaindu

cible

dom

ainfamily,mem

ber

2A4.06

861.27

4.01

5q35

.2

4085

0_at

FKBP8

FK50

6bindingprotein

8,38

kDa

4.07

681.35

4.01

19p12

2035

07_a

tCD68

CD68

molec

ule

4.08

091.22

4.01

17p13

2016

05_x

_at

CNN2

Calpon

in2

4.10

171.59

4.01

19p13

.321

1793

_s_a

tABI2

Abl-interactor

24.10

521.26

4.01

2q33

2178

06_s

_at

POLD

IP2

Polym

eras

e(DNA-dire

cted

),delta

interactingprotein

24.13

881.26

3.26

17q11

.2

2212

69_s

_at

SH3B

GRL3

SH3dom

ainbind

ingglutam

icac

id-richprotein

like3

4.19

761.58

3.26

1p35

–p34

.3

2116

22_s

_at

ARF3

ADP-ribos

ylationfactor

34.20

201.25

3.26

12q13

2174

57_s

_at

RAP1G

DS1

RAP1,

GTP

-GDPdisso

ciation

stim

ulator

14.22

391.26

1.95

4q23

–q25

2048

75_s

_at

GMDS

GDP-m

anno

se4,6-deh

ydratase

4.24

551.63

1.95

6p25

2177

36_s

_at

EIF2A

K1

Euk

aryo

tictran

slation

initiationfactor

2-alph

akina

se1

4.27

281.57

1.95

7p22

2057

70_a

tGSR

Glutathione

reduc

tase

4.28

001.35

1.95

8p21

.120

7722

_s_a

tBTB

D2

BTB

(POZ)do

mainco

ntaining

24.30

071.50

1.95

19p13

.320

6200

_s_a

tANXA11

Ann

exin

A11

4.30

521.62

1.95

10q23

2088

77_a

tPAK2

p21

protein(Cdc4

2/Rac

)-ac

tivated

kina

se2

4.34

491.84

1.95

3q29

2125

66_a

tMAP4

Microtubule-as

sociated

protein

44.36

081.50

1.95

3p21

2007

88_s

_at

PEA15

Pho

spho

protein

enric

hed

inas

troc

ytes

154.38

381.57

1.95

1q21

.1

2191

62_s

_at

MRPL1

1Mito

chon

dria

lribos

omal

protein

L11

4.42

121.29

1.95

11q13

.3

2172

11_a

t—

—4.44

141.41

1.95

—

3220

9_at

FAM89

BFa

mily

with

seque

nce

similarity89

,mem

ber

B4.46

151.70

1.95

11q23

2123

30_a

tTF

DP1

Tran

scrip

tionfactor

Dp-1

4.50

851.95

1.95

13q34

2020

41_s

_at

FIBP

Fibroblast

grow

thfactor

(acidic)

intrac

ellularbindingprotein

4.51

751.71

1.95

11q13

.1

2086

77_s

_at

BSG

Bas

igin

(Okblood

grou

p)

4.55

871.78

1.95

19p13

.3

(Con

tinue

don

thefollo

wingpag

e)






Tab

le2.

Func

tiona

lan

notatio

nsof

the

63ge

nes

iden

tified

asdifferen

tially

expressed

amon

gthe

2NMF

grou

ps

by

SAM

2-clas

san

alysis

(Con

t'd)

Probese

tID

Gen

esy

mbol

Gen

etitle

Sco

re(d)

Fold

chan

ge

q(%

)Cytoban

d

2000

01_a

tCAPNS1

Calpain,

smalls

ubun

it1

4.61

761.71

019

q13

.12

2180

37_a

tFA

M13

4AFa

mily

with

seque

ncesimilarity

134,

mem

ber

A4.70

821.48

02q

35

3796

5_at

PARVB

Parvin,

beta

4.78

961.48

022

q13

.2–q13

.33

2137

39_a

t—

—4.81

741.28

0—

2210

59_s_a

tCOTL

1Coa

ctos

in-like1(Dictyos

telium)

4.85

242.20

016

q24

.121

1672

_s_a

tARPC4///TT

LL3

Actin-related

protein2/3

complex,

subun

it4,

20kD

a///

tubulin

tyrosine

ligas

e-likefamily,mem

ber

3

4.90

821.68

03p

25.3

2008

38_a

tCTS

BCathe

psinB

5.09

361.58

08p

2220

1426

_s_a

tVIM

Vim

entin

5.12

721.82

010

p13

2134

53_x

_at

GAPDH

Glyce

raldeh

yde-3-ph

ospha

tedeh

ydroge

nase

5.29

861.70

012

p13

2141

19_s_a

tFK

BP1A

FK50

6binding

protein1A

,12

kDa

5.32

261.78

020

p13

2090

41_s_a

tUBE2G

2Ubiquitin

-con

juga

tingen

zyme

E2G

2(UBC7ho

molog

,ye

ast)

5.38

331.65

021

q22

.3

2012

51_a

tPKM2

Pyruv

atekina

se,mus

cle

5.41

681.92

015

q22

2088

29_a

tTA

PBP

TAPbindingprotein

(tapas

in)

5.52

711.55

06p

21.3

2010

82_s_a

tDCTN

1Dyn

actin

1(p15

0,glue

dho

molog

,Droso

phila)

5.53

972.09

02p

13

2183

88_a

tPGLS

6-Pho

spho

gluc

onolac

tona

se5.88

841.89

019

p13

.220

1516

_at

SRM

Spermidinesy

ntha

se5.90

121.57

01p

36–p22

2011

36_a

tPLP

2Proteolipid

protein

2(colon

icep

ithelium-enriche

d)

6.04

221.62

0Xp11

.23

2009

66_x

_at

ALD

OA

Aldolas

eA,fruc

tose

-bispho

spha

te6.10

411.72

016

p11

.220

3683

_s_a

tVEGFB

Vas

cularen

dothe

lialg

rowth

factor

B6.33

301.51

011

q13

2083

08_s_a

tGPI

Gluco

sepho

spha

teisom

eras

e7.12

611.94

019

q13

.1

Mosca et al.





The natural grouping of genome profiles by SNP arrayanalysis showed that the complex scenario of CN altera-tions affecting CLL is mainly driven by the presence of the13q14 deletion and trisomy 12. Our SNP array approachextends previous limited evidence showing that del(13)(q14) patients are characterized by differently sized dele-tions. Ouillette et al. (9) documented a significant correla-tion between a wider deletion encompassing the RB1 geneand a higher Rai clinical stage at diagnosis or in previouslytreated patients. In preliminary observations on our cohortof Binet stage A patients, we did not find any significantdifferences in the time to treatment between the patientswith shorter/biallelic (group I) or wider/monoallelic losses(group II; data not shown), although this finding awaitsconfirmation by ongoing prospective studies.

Our data support the notion that loss of the miR-15a/16-1 locus occurs in virtually all the CLL cases with del(13)(q14); in fact, among 44 cases with 13q deletion in ourstudy, both copies of miRNA cluster were retained in only 2patients with a monoallelic 13q14 deletion. Furthermore,we found that reduced miR-15a and miR-16 expressionlevels in patients with del(13)(q14) significantly correlatedonly with the presence of a biallelic loss, which is inagreement with some recent data (8, 9) but not with theolder findings (7). Overall, these findings suggest a need toredefine the pathogenetic role of miR-15a and miR-16-1 inthe context of the molecular subtypes of 13q14-deletedpatients.

The availability and integration of GEP data allowed amore comprehensive overview of the genetic complexity ofthe 13q14 deletion. Indeed, this approach led to thedetermination of distinct transcriptional signatures asso-ciated with different groups of 13q14-deleted patients(short/biallelic versus long/monoallelic lesion) identifiedby the NMF algorithm. In particular, a significant genedosage effect has been observed involving the downregula-tion of genes in group II (long/monoallelic deletion), 6 ofwhich localized within the 13q14 region. Among thedownregulated genes, we should note TPT1/TCTP (trans-lationally controlled tumor protein) encoding for a highlyconserved multifunctional protein acting as a prosurvivaland growth stimulating factor (24), which inhibits BAX-induced apoptosis (25).

Most of the upregulated genes in group II are involved incell motility and adhesion, regulation of cell proliferation,tumor cell migration, metastasis, angiogenesis, and apop-tosis, and some of these may contribute to lymphomagen-esis. This is the case of the autocrine motility factor (AMF)/glucose phosphate isomerase (GPI) gene, which is upre-gulated in several human cancers and encodes for a house-keeping cytosolic enzyme involved in both glycolysis andgluconeogenesis (26). The basigin gene (BSG) encodes acell surface glycoprotein of the Ig superfamily expressed atthe surface of tumor cells metastasizing in bone marrow(27) and is believed to induce matrix metalloproteinasesproduction (28). The Galectin-1 (alias LGALS1) geneencoding a 14-kDa lectin, is overexpressed in numeroustumors including lymphomas (29) and CLL, and is

Fig. 4. Identification of gene signatures characterizing 13q14 classes.A, dendrogram of the 22 CLL samples clustered according to theexpression profiles of the genes located at 13q14. B, expression profiles ofthe NMF CLL group I versus group II for the 76 probe sets selected by aSAM 2-class analysis. Information on chromosomes 11, 13, and 17deletions, chromosome 12 trisomy (þ, positive; –, negative; /, biallelicdeletion), CD38 (þ, �30%; –, <30%), ZAP-70 expression (þ, �30%; –,<30%), and IgVH mutational status (þ, mutated; –, unmutated) is includedalongside the patient ID. The color scale bar represents the relativechanges in gene expression normalized by the standard deviation, and thecolor changes in each row represent gene expression in relation to themean across the samples.






implicated in abnormal mechanisms of cell adhesion,induction of apoptosis, and tumor angiogenesis (30,31). PAK2 belongs to the p21-activated kinase family,which are well-known regulators of cytoskeletal remodel-ling, cell motility, proliferation, and apoptosis (32).Finally, 2 other upregulated genes, b-parvin (PARVB)and vimentin (VIM), have been reported to play a criticalrole in transducing signals from integrins to the actincytoskeleton and intracellular signaling proteins (33).Low levels of PARVB correlate with low adhesion to col-lagen (34), whereas increased levels reduce the activatingphosphorylation of AKT (35), causing propensity to apop-tosis. Upregulation of VIM is thought to provide a selectiveadvantage to tumor cells following signaling cues frommesenchymal and epithelial extracellular matrixes (36).Notably, a modulation of genes thought to act as regulatorsof tumor invasion and integrin-mediated cell motility andadhesion has been recently described in CLL, in particularduring disease progression (37, 38).

In conclusion, our data may represent a valid contribu-tion to the definition of the genomic profile of CLL. Inparticular, we provide evidence of 2 clearly distinguishable

molecular subtypes among CLL patients with 13q14 dele-tion thatmay contribute toward the better understanding ofthe pathogenetic and clinical relevance of this lesion in CLL.

Disclosure of Potential Conflicts of Interest

No potential conflict of interest was disclosed.

Grant Support

This study was supported by grants from AIRC to A.N. (IG 4659)and F.M. (RG 6432 cofinanced by AIRC, CARICAL, Fondazione ‘AmeliaScorza’ and Provincia di Cosenza); AIL Sezione Milano; Fondazione‘Amelia Scorza’ ONLUS, Cosenza; Progetti Strategici–Ricerca FinalizzataMinistero Italiano della Salute "RFPS20063339960" (to G.C.) and"RFPS2006340196" (to F.M. and M.F.); and FIRB (Grant RBIP06LCA9 toM.F.). The Progetto Ordinario Ricerca Finalizzata Ministero Italianodella Salute-2007 (to G.C.), Progetto Compagnia San Paolo (to G.C.),and the Fondazione Internazionale Ricerche Medicina Sperimentale(FIRMS) provided financial and administrative assistance. L.A. and S.M.were supported by fellowships from the Fondazione Italiana Ricerca sulCancro (FIRC).

Received 01/19/2010; revised 06/25/2010; accepted 07/24/2010;published OnlineFirst 10/14/2010.

References1. Chiorazzi N, Rai KR, Ferrarini M. Chronic lymphocytic leukemia. N

Engl J Med 2005;352:804–15.2. Crespo M, Bosch F, Villamor N, Bellosillo B, Colomer D, Rozman M,

et al. ZAP-70 expression as a surrogate for immunoglobulin-variable-region mutations in chronic lymphocytic leukemia. N Engl J Med2003;348:1764–75.

3. Damle RN, Wasil T, Fais F, Ghiotto F, Valetto A, Allen SL, et al. Ig Vgene mutation status and CD38 expression as novel prognosticindicators in chronic lymphocytic leukemia. Blood 1999;94:1840–7.

4. Dohner H, Stilgenbauer S, Benner A, Leupolt E, Kr€ober A, Bullinger L,et al. Genomic aberrations and survival in chronic lymphocytic leu-kemia. N Engl J Med 2000;343:1910–6.

5. Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK.Unmutated Ig V(H) genes are associated with a more aggressive formof chronic lymphocytic leukemia. Blood 1999;94:1848–54.

6. M�ehes G. Chromosome abnormalities with prognostic impact in B-cell chronic lymphocytic leukemia. Pathol Oncol Res 2005;11:205–10.

7. Calin GA, Dumitru CD, Shimizu M, Bichi R, Zupo S, Noch E, et al.Frequent deletions and down-regulation of micro- RNA genes miR15and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl AcadSci U S A 2002;99:15524–9.

8. Fulci V, Chiaretti S, Goldoni M, Azzalin G, Carucci N, Tavolaro S, et al.Quantitative technologies establish a novel microRNA profile ofchronic lymphocytic leukemia. Blood 2007;109:4944–51.

9. Ouillette P, Erba H, Kujawski L, Kaminski M, Shedden K, Malek SN.Integrated genomic profiling of chronic lymphocytic leukemia identi-fies subtypes of deletion 13q14. Cancer Res 2008;68:1012–21.

10. Morabito F, Cutrona G, Gentile M, Matis S, Todoerti K, Colombo M,et al. Definition of progression risk based on combinations of cellularand molecular markers in patients with Binet stage A chronic lym-phocytic leukaemia.Br J Haematol 2009;146:44–53.

11. Cutrona G, Colombo M, Matis S, Reverberi D, Dono M, Tarantino V,et al. B lymphocytes in humans express ZAP-70 when activated invivo.Eur J Immunol 2006;36:558–69.

12. Fais F, Ghiotto F, Hashimoto S, Sellars B, Valetto A, Allen SL, et al.Chronic lymphocytic leukemia B cells express restricted sets ofmutated and unmutated antigen receptors. J Clin Invest 1998;102:1515–25.

13. Fabris S, Mosca L, Todoerti K, Cutrona G, Lionetti M, Intini D, et al.Molecular and transcriptional characterization of 17p loss in B-cell

chronic lymphocytic leukemia. Genes Chromosomes Cancer 2008;47:781–93.

14. Agnelli L, Mosca L, Fabris S, Lionetti M, Andronache A, Kwee I, et al. ASNP microarray and FISH-based procedure to detect allelic imbal-ances in multiple myeloma: an integrated genomics approach reveals awide gene dosage effect. GenesChromosomesCancer 2009;48:603–14.

15. Olshen AB, Venkatraman ES, Lucito R, Wigler M. Circular binarysegmentation for the analysis of array-based DNA copy number data.Biostatistics 2004;5:557–72.

16. Brunet JP, Tamayo P, Golub TR, Mesirov JP. Metagenes and mole-cular pattern discovery using matrix factorization. Proc Natl Acad SciU S A 2004;101:4164–9.

17. Heid CA, Stevens J, Livak KJ, Williams PM. Real time quantitativePCR.Genome Res 1996;6:986–94.

18. Bodin L, Beaune PH, Loriot MA. Determination of cytochrome P4502D6 (CYP2D6) gene copy number by real-time quantitative PCR. JBiomed Biotechnol 2005;2005:248–53.

19. Lionetti M, Agnelli L, Mosca L, Fabris S, Andronache A, Todoerti K,et al. Integrative high-resolution microarray analysis of human mye-loma cell lines reveals deregulated miRNA expression associated withallelic imbalances and gene expression profiles.Genes ChromosomesCancer 2009;48:521–31.

20. Agnelli L, Bicciato S, Mattioli M, Fabris S, Intini D, Verdelli D, et al.Molecular classification of multiple myeloma: a distinct transcriptionalprofile characterizes patients expressing CCND1 and negative for14q32 translocations. J Clin Oncol 2005;23:7296–306.

21. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarraysapplied to the ionizing radiation response. Proc Natl Acad Sci U S A2001;98:5116–21.

22. Schadt EE, Li C, Ellis B, Wong WH. Feature extraction and normal-ization algorithms for high-density oligonucleotide gene expressionarray data. J Cell Biochem Suppl 2001;Suppl37:120–5.

23. Pfeifer D, Pantic M, Skatulla I, Rawluk J, Kreutz C, Martens UM, et al.Genome-wide analysis of DNA copy number changes and LOH in CLLusing high-density SNP arrays. Blood 2007;109:1202–10.

24. Telerman A, Amson R. Themolecular programme of tumour reversion:the steps beyond malignant transformation. Nat Rev Cancer 2009;9:206–16.

25. Li F, Zhang D, Fujise K. Characterization of fortilin, a novel antiapop-totic protein. J Biol Chem 2001;276:47542–9.

Mosca et al.





26. Harrison RA. The detection of hexokinase, glucosephosphate iso-merase and phosphoglucomutase activities in polyacrylamide gelsafter electrophoresis: a novel method using immobilized glucose 6-phosphate dehydrogenase. Anal Biochem 1974;61:500–7.

27. Nabeshima K, Iwasaki H, Koga K, Hojo H, Suzumiya J, Kikuchi M.Emmprin (basigin/CD147): matrix metalloproteinase modulator andmultifunctional cell recognition molecule that plays a critical role incancer progression. Pathol Int 2006;56:359–67.

28. Biswas C. Tumor cell stimulation of collagenase production by fibro-blasts. Biochem Biophys Res Commun 1982;109:1026–34.

29. Demydenko D, Berest I. Expression of galectin-1 in malignant tumors.Exp Oncol 2009;31:74–9.

30. Perillo NL, Pace KE, Seilhamer JJ, Baum LG. Apoptosis of T cellsmediated by galectin-1. Nature 1995;378:736–9.

31. Thijssen VL, Postel R, Brandwijk RJ, Dings RP, Nesmelova I, Satijn S,et al. Galectin-1 is essential in tumor angiogenesis and is a targetfor antiangiogenesis therapy. Proc Natl Acad Sci U S A 2006;103:15975–80.

32. Molli PR, Li DQ, Murray BW, Rayala SK, Kumar R. PAK signaling inoncogenesis. Oncogene 2009;28:2545–55.

33. Sepulveda JL, Wu C. The parvins. Cell Mol Life Sci 2006;63:25–35.34. Mongroo PS, Johnstone CN, Naruszewicz I, Leung-Hagesteijn C,

Sung RK, Carnio L, et al. Beta-parvin inhibits integrin-linked kinasesignaling and is downregulated in breast cancer. Oncogene2004;23:8959–70.

35. Zhang Y, Chen K, Tu Y, Wu C. Distinct roles of two structurally closelyrelated focal adhesion proteins, alpha-parvins and beta-parvins, inregulation of cell morphology and survival. J Biol Chem2004;279:41695–705.

36. Hendrix MJ, Seftor EA, Chu YW, Trevor KT, Seftor RE. Role ofintermediate filaments in migration, invasion and metastasis. CancerMetastasis Rev 1996;15:507–25.

37. Fernandez V, Jares P, Salaverria I, Gin�e E, Be�a S, Aymerich M, et al.Gene expression profile and genomic changes in disease progressionof early-stage chronic lymphocytic leukemia. Haematologica2008;93:132–6.

38. Haran M, Chebatco S, Flaishon L, Lantner F, Harpaz N, Valinsky L,et al. Grb7 expression and cellular migration in chronic lymphocyticleukemia: a comparative study of early and advanced stage disease.Leukemia 2004;18:1948–50.






2010;16:5641-5653. Published OnlineFirst October 14, 2010.Clin Cancer Res Laura Mosca, Sonia Fabris, Marta Lionetti, et al. with 13q14 DeletionSubgroups of B-Cell Chronic Lymphocytic Leukemia Patients Integrative Genomics Analyses Reveal Molecularly Distinct

Updated version

10.1158/1078-0432.CCR-10-0151doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2010/12/08/1078-0432.CCR-10-0151.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/16/23/5641.full#ref-list-1

This article cites 38 articles, 14 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/16/23/5641.full#related-urls

This article has been cited by 7 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

SubscriptionsReprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/16/23/5641To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-10-0151

http://clincancerres.aacrjournals.org/content/suppl/2010/12/08/1078-0432.CCR-10-0151.DC1

http://clincancerres.aacrjournals.org/content/16/23/5641.full#ref-list-1

http://clincancerres.aacrjournals.org/content/16/23/5641.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/16/23/5641


Documents

Integrative Genomics Analyses Reveal Molecularly Distinct ... · Integrative Genomics Analyses Reveal Molecularly Distinct Subgroups of B-Cell Chronic Lymphocytic Leukemia Patients