Translocation in Myeloma

5692 Vol. 10, 5692–5701, September 1, 2004 Clinical Cancer Research

Featured Article A Global Expression-based Analysis of the Consequences of the t(4;14) Translocation in Myeloma

Ann M. Dring,1 Faith E. Davies,1 Other important functional classes of genes include cell James A. L. Fenton,1 Philippa L. Roddam,1 signaling, apoptosis and related genes, oncogenes, chromatin Kathryn Scott,1 David Gonzalez,1 Sara Rollinson,1 structure, and DNA repair genes. Interestingly, 25% of 1 1 myeloma cases lacking evidence of this translocation had Andrew C. Rawstron, Karen S. Rees-Unwin, up-regulation of the MMSET transcript to the same level as 2 3 Cheng Li, Nikhil C. Munshi, cases with a translocation. Kenneth C. Anderson,3 and Gareth J. Morgan4 Conclusions: t(4;14) cases form a distinct subgroup of 1Academic Unit of Haematology and Oncology, University of Leeds, myeloma cases with a unique gene signature that may ac- Leeds, United Kingdom; 2Department of Biostatistics, Harvard School count for their poor prognosis. A number of non-t(4;14) 3 of Public Health, Boston, Massachusetts; Jerome Lipper Multiple cases also express MMSET consistent with this gene playing Myeloma Center; Dana-Farber Cancer Institute, Boston, a role in myeloma pathogenesis. Massachusetts; and 4Royal Marsden Hospital, Surrey, United Kingdom INTRODUCTION ϳ ABSTRACT The t(4;14) is seen in 10–20% of presenting cases of multiple myeloma and has been suggested to be associated with Purpose: Our purpose in this report was to define genes a poor prognosis (1–4). The molecular events leading to the and pathways dysregulated as a consequence of the t(4;14) translocation have been extensively studied and are thought to in myeloma, and to gain insight into the downstream func- be due to aberrant immunoglobulin class switching recombina- tional effects that may explain the different prognosis of this tion (5–10). As a consequence of the rearrangement, there is a subgroup. reciprocal translocation of genetic material between the immu- Experimental Design: Fibroblast growth factor receptor 3 noglobulin heavy chain region (IgH) on chromosome 14q32 and (FGFR3) overexpression, the presence of immunoglobulin the 5Ј region of MMSET on chromosome 4p16.3. This leads to heavy chain-multiple myeloma SET domain (IgH-MMSET) at least two genes, fibroblast growth factor receptor 3 (FGFR3) fusion products and the identification of t(4;14) breakpoints on the der(14) and multiple myeloma with a SET domain were determined in a series of myeloma cases. Differentially (MMSET) on the der(4), being brought under the influence of (5 ؍ expressed genes were identified between cases with (n the strong immunoglobulin gene enhancers E␣ and E␮ respec- -a t(4;14) by using global gene expres (24 ؍ and without (n tively. The molecular characterization of cell lines known to sion analysis. carry the t(4;14) have identified the consistent up-regulation of Results: Cases with a t(4;14) have a distinct expression FGFR3 expression and the formation of a novel hybrid RNA pattern compared with other cases of myeloma. A total of fusion product consisting of upstream immunoglobulin se- 127 genes were identified as being differentially expressed quences fused to the downstream sequences of MMSET (IgH- including MMSET and cyclin D2, which have been previ- MMSET). ously reported as being associated with this translocation. These cell line observations have been extended into patient material in which we and others have demonstrated that FGFR3 is overexpressed in the majority of cases carrying a Received 3/8/04; revised 5/20/04; accepted 6/2/04. t(4;14) translocation (9, 11). However, although an IgH-MMSET Grant support: Supported by the Leukaemia Research Fund United hybrid transcript can be detected in all cases with FGFR3 Kingdom (F. Davies, G. Morgan), Department of Health (F. Davies, overexpression, the reverse is not consistently so (12). Cases K. Rees-Unwin), Yorkshire Cancer Research (D. Gonzalez, G. Mor- have been identified where IgH/MMSET hybrid transcripts are gan), British Society of Haematology (A. Rawstrom, F. Davies), and Doris Duke Distinguished Clinical Research Scientist Award present but FGFR3 overexpression is lacking because of an (K. Anderson). interstitial deletion within FGFR3. This has lead to the sugges- The costs of publication of this article were defrayed in part by the tion that, although FGFR3 deregulation may be critical for the payment of page charges. This article must therefore be hereby marked early stages of myeloma pathogenesis, dependence on it may be advertisement in accordance with 18 U.S.C. Section 1734 solely to lost later in the disease course. indicate this fact. Note: Supplementary data for this article are available at Clinical In contrast to FGFR3, the deregulation of MMSET seems to Cancer Research Online (http://clincancerres.aacrjournals.org); A. be a universal feature of cases carrying a t(4;14). The site of the Dring and F. Davies contributed equally to this work. breakpoints on chromosome 4 have been well described, and the Requests for reprints: Faith Davies, Academic Unit of Haematology location of these breaks suggest that cases with a t(4;14) can be and Oncology, Algernon Firth Building, School of Medicine, University of Leeds, Leeds, United Kingdom LS2 9JT. Phone/Fax: 44-113-343- further subdivided according to the variant of MMSET they 3372; E-mail: [email protected]. express. Studies of the coding sequence have identified two ©2004 American Association for Cancer Research. initiation sites located in exon 3 and exon 15, leading to the

production of three possible transcripts (7). Transcripts initiated Gene Array. RNA was extracted by using commer- from exon 3 are polyadenylated in either exon 11 (type I), or cially available kits (Qiagen and Stratagene, La Jolla, CA), exon 24 (type II) as a result of alternative splicing occurring according to manufacturers’ instructions, and was amplified from exon 10 to 11, or exon 10 to 12. A further possible with a modified SMART PCR protocol (BD Biosciences, transcript, type III (IREII-BP), may also be produced by initia- Palo Alto, CA), in which a 5Ј T7 polymerase promoter site tion from exon 15. Breakpoints in exon 2 are able to produce was incorporated to create amplified cDNA compatible with three possible transcripts (type I 647 aa, type II 1365 aa, and downstream processing for the Affymetrix GeneChip system type III 584 aa), whereas breaks in or after exon 3 can only (14). Human Genome U95Av2 GeneChip arrays (Affymetrix, produce one transcript (type III 584 aa). Santa Clara, CA) containing probes for 12,600 expressed To investigate the consequences of the t(4;14), we chose to sequences, were used for mRNA expression profiling, as use a global expression microarray approach to look at the described previously (14). pattern of genes expressed in plasma cells isolated from cases of Array Analysis. Array normalization and expression myeloma that constitutively express FGFR3, compared with value calculation was performed by using DNA-Chip Analyzer cases of myeloma that do not. The aim being to discover genes (dChip; freely available to academic users at www.dchip.org; 5 and pathways dysregulated as a consequence of the t(4;14) and ref. 15). Invariant set normalization was used to normalize to gain insight into the downstream functional effects occurring arrays at the probe cell level, and the perfect-match only model- in these cells that differentiate them from other cases of my- based method (15, 16) was used for outlier detection, probe eloma and that could explain the different prognosis of this selection, and computing expression values. The 29 myeloma patient chips were run at four different time points, which made subgroup. it necessary to pre-process the data to minimize the number of genes identified as different due to batch differences. In dChip, MATERIALS AND METHODS an array list file with “standardize separators” was used to Samples and Plasma Cell Selection. Bone marrow as- separate the chips into their respective batches to generate pirate samples were obtained after informed consent from 29 gene-wise standardized values, which were then exported into a myeloma patients. After 0.86% ammonium chloride red cell text file. These batch standardized data were then read into lysis, plasma cells were positively selected by using CD138 dChip using the “Get external data” function. It was then pos- microbeads and Magnet Assisted Cell Sorting (Miltenyi Bio- sible to use the “Compare Samples” function to identify genes tech, Bergisch Gladbach, Germany), according to the manufac- differentially expressed between myeloma cases with and without a t(4;14) translocation. Cases without a t(4;14) were speci- turer’s instructions. The final purity of these cells (Ͼ95%) was fied as the baseline group (B), and cases with a t(4;14) the assessed by morphology and flow cytometry (FACSort BD experimental group (E). As a result of the standardization pro- Biosciences, San Jose, CA). cedure it was not possible to compute traditional fold changes; Detection of the t(4;14) by Reverse Transcription-PCR therefore, the standardized expression values in the E and B and Characterization of IgH/MMSET Breakpoints. Pa- groups were used for unpaired two-sample t tests. Hierarchical tients with t(4;14) translocations were identified by a combina- clustering was performed on the original expression data, using tion of two reverse transcription-polymerase chain reaction (RT- gene lists resulting from “compare samples,” and “standardize PCR) assays as described previously (9). The first approach separators” to handle any batch-specific effects. An unsuper- used PCR to detect the IgH-MMSET fusion product, with prim- vised analysis was also carried out by using the dChip “filter ers that span the breakpoint (13), and the second demonstrated genes” function, for which the variation across samples was as overexpression of the FGFR3 transcript, which is only seen in follows: 1.10 Ͻ SD/mean Ͻ 10.00, and Present call % in the the presence of a translocation (9). cDNA was prepared by using array used was Ն20%. the Superscript reverse transcriptase enzyme (Invitrogen Life Technologies, Inc. Paisley, United Kingdom) with random hex- RESULTS amer primers. PCR primer sequences are reported elsewhere (9, FGFR3 Expression and Splice Variants. Five cases 13). Both PCR assays used LA Taq polymerase (Takara Bio were identified by PCR as having a t(4;14) because they ex- Inc., Shiga, Japan) and the following thermal cycling conditions: pressed the IgH-MMSET fusion product together with overex- 95°C for 3 minutes; 35 cycles of 95°C for 30 seconds, 60°C for pression of FGFR3. Patients with detectable FGFR3 expressed 45 seconds, and 72°C for 45 seconds; ending at 72°C for 5 mRNA for both the FGFR3 IIIb and FGFR3 IIIc isoforms, minutes. For sequencing, DNA bands were separated with 1.5% resulting from alternative splicing in the ligand-binding domain. agarose gel electrophoresis and were stained with ethidium FGFR3 mutation analysis, looking for the common mutations of bromide. PCR products were excised and cleaned by using the these cases, revealed no evidence of mutations. For each case, QIAquick gel purification system (Qiagen, Crawley, UK) ac- the location of the IgH-MMSET splice site was identified; the cording to the manufacturer’s instructions. DNA bands were genomic breakpoints were located 5Ј of these sites (Fig. 1; ref. sequenced in both directions by using PCR primers. Approxi- 10). The remaining 24 cases were negative for IgH-MMSET mately 100 ng of purified DNA were used in the sequencing transcripts and FGFR3 overexpression. reaction (Big Dye Terminator kit; PE Biosystems, Warrington, United Kingdom); and the sequences were analyzed with sequence analysis software (Applied Biosystems, Warrington, UK). 5 Internet address for DNA-Chip Analyzer (dChip): www.dchip.org.

most across the multiple myeloma patient group. When used as the basis for hierarchical clustering, four out of five of t(4;14) myelomas were placed in the same main branch of the dendro- gram (P Ͻ 0.0357; see supplementary information. These genes included cyclin D1 and dickkopf homolog 1, which have been previously identified as being differentially expressed in myeloma (17, 18). Two genes (cyclin D2 and amphiregulin) were common to the list of genes identified in the supervised analysis as being associated with the t(4;14). Molecular Consequences for Genes on Chromosome 4. The IgH enhancer regions are strong and can act over many kilobases of DNA to affect gene expression, potentially deregu- lating genes not directly at the site of the translocation. We used the expression data derived from the arrays to look at genes on 4p that could be potentially deregulated by the translocation (Table 1). The only gene located directly at the breakpoint to be deregulated was MMSET, which was up-regulated in all of the t(4;14) cases, confirming prior data from cell lines. Two other genes, PHOX2B and ARHH, located on 4p12 and 4p13, respectively, were also significantly different between cases with and without the translocation. However, these genes are located many hundreds of kilobases from the translocation breakpoints and may not be deregulated directly by the translocation. It is important to note that a number of genes located around the breakpoint, including telomeric SPON2 and TACC3 and centro- meric LETM1 and TNIP2, are not represented on the Affymetrix U95Av2 chip. Fig. 1 In A, RT-PCR detection of the IgH-MMSET fusion transcript in MMSET Overexpression Is Not Restricted to Cases six multiple myeloma patients (Lanes 2 to 6) and the cell lines H929 and with a t(4;14). Interestingly, 6 (25%) of the 24 cases of KMS11 (Lanes 7 and 8, respectively) demonstrates three different myeloma lacking evidence of a translocation by RT-PCR did product sizes depending on which exons of MMSET are present. Lanes 3–5 show IgH-MMSET negative patients. Lane 1 is 100 bp ladder. B have up-regulation of an MMSET transcript to the same level as demonstrates diagramatically that the 218-bp band of Lane 2 comprises cases with a translocation (Fig. 3). To examine these cases exons 5 to 6, Lanes 6 and KMS11 have a 1,025-bp product comprising all of the exons from 3 to 6, and H929 has a 381-bp band containing exons 4, 5, and 6. C, example of mRNA sequence from Lane 2 patient showing the position of the splice site. Bold, primer sequences. Under- lined base is polymorphic

Multiple Myeloma Cases Carrying the t(4;14) Have a Distinct Gene Expression Profile. Cases with a t(4;14) have a distinct expression pattern compared with other cases of myeloma, and, using the analysis criteria outlined above, we identified 127 genes as differentially expressed between myeloma plasma cells with and without a t(4;14) (see Supplemen- tary Information for complete gene list).6 Clustering using this gene list and the original expression values confirmed that the selected genes could separate t(4;14) from non-t(4;14) cases (Fig. 2). Of these genes, 67 were overexpressed in the t(4;14) cases compared with non-t(4;14) cases, and 60 were underexpressed (Fig. 2). Importantly, some genes included in this list, e.g., MMSET and cyclin D2, have been previously reported as being associated with the t(4;14) translocation. An additional unsupervised analysis, performed with the dChip “filter genes” function, identified 29 genes that varied

Fig. 2 Supervised hierarchical clustering of t(4;14) and non-t(4;14) myeloma patients with the use of genes identified as differentially 6 See Supplementary Data for this article at http://cancerres.aacrjournals.org. expressed between the two subgroups.

Table 1 Genes with significant mean differences between t(4;14) and non-t(4;14) cases located on chromosomes 4p and 14q Gene Name Chromosome Accession no. Mean difference P value PHOX2B Paired-like homeobox 2B 4p12 D82344 Ϫ1.07 0.003 ARHH Ras homolog gene family, member H 4p13 Z35227 1.19 0.04 WHSC1 Wolf-Hirschorn syndrome candidate 1 (MMSET) 4p16.3 AJ007042 1.27 0.005 ZFP36L1 Zinc finger protein 36, C3H type-like 1 14q22–14q24 X79067 1.44 0.010 TMP21 Transmembrane trafficking protein 14q24.3 L40397 1.25 0.024 L40391 1.19 0.005 CALM1 Calmodulin 1 (phosphorylase kinase, ␦) 14q24-q31 U16850 1.16 0.020 EIF5 Eukaryotic translation initiation factor 5 14q32.33 AL080102 1.15 0.027

further, we looked at the pattern of hybridization of the individ- as being differentially expressed between cases with and with- ual oligonucleotide sequences representing the MMSET gene out a t(4;14) translocation. Down-regulated genes included arrayed onto the Affymetrix chip. The gene is represented by 16 Caspase 2 apoptosis-related cysteine protease (CASP2), the oligonucleotides that are all situated in the 3Ј region of the gene active form of which is induced by a variety of apoptotic stimuli, in exon 24 (Fig. 4). The hybridization pattern seen in cases with and which a recent report demonstrated is also decreased in a t(4;14) as well as in cases with an increase in MMSET levels mantle cell lymphoma (19); Histone deacetylase 1 (HDAC1), in the absence of a translocation was similar and covered the which encodes for a component of the histone deacetylase whole of the probe set. This confirmed that the expression complex, which interacts with retinoblastoma tumor suppressor changes are real and are not due to the cross-hybridization of a protein and is a key element in the control of cell proliferation few bad probes. and differentiation; Death effector domain containing gene Functional Classes of Genes Differentially Expressed (DEDD), which regulates programmed cell death and connects between t(4;14) and Non-t(4;14) Cases. The underlying con- with the caspases to mediate formation of the death-inducing sequences of the translocation on the biology of malignant signal complex; and Immediate early response gene X-1 (IER3/ plasma cells may be understood more clearly by studying the IEX1) which is regulated by multiple transcription factors functional classes of the genes altered. The results of this anal- including p53, NF-␬B/rel, Sp1, and c-Myc and which protects ysis are presented in Table 2. For each gene the Human Genome cells from Fas- or tumor necrosis factor type ␣-induced Organization (HUGO) gene nomenclature committee-approved apoptosis. gene name is used, when possible, followed by the abbreviation Membrane Genes. A number of key adhesion molecules in parentheses. were noted to be up-regulated in the t(4;14) cases. These include Cell Signaling. A number of RAS family genes distin- N-Cadherin (CDH2), which is both an adhesion molecule and guished the t(4;14) cases from cases not carrying a translocation. an important downstream target of FGFR3 signaling; Activated These include: an up-regulation of RAS guanyl releasing protein leukocyte cell adhesion molecule (ALCAM/CD166), a member 1 (RAS GRP1) an activator of RAS signaling, and ras homolog of the immunoglobulin superfamily, which mediates both het- gene family members Q and H (ARHQ and ARHH). A number erophilic (ALCAM-CD6) and homophilic (ALCAM-ALCAM) of members are also down-regulated, including Rap2 binding cell–cell interactions; Vimentin, which has been previously protein 9, and mitogen-activating protein kinase 14 (MAPK14), linked with an invasive phenotype of prostate cancer and epi- which has been linked with tumor progression. Amphiregulin thelial carcinomas; Decorin, an important component of the (AREG), an epidermal growth factor family member, was up- proteoglycan core, which binds type I collagen; and CD43 regulated in t(4;14) cases. This gene has been reported to be an autocrine growth factor in other systems and may act as a survival factor via the RAS/RAF/MAPK pathway. A number of FGF-related genes were noted to be down-regulated in t(4;14) cases. These include FGF receptor activating protein 1 (FRAG1), which has been identified as an oncogenic event in osteosarcoma because of a chromosomal rearrangement between fibroblast growth factor receptor 2 (FGFR2) and FRAG1; and glycoprotein A repetitions predominant (GARP), which encodes a type I membrane protein that has been linked with hormone escape in prostate cancer and ovarian cancer. The final common mechanism underlying dysregulated cellular signaling is the alteration of transcription within the nucleus. Genes that may play a role in this include a number of zinc finger proteins (ZFP36L1 and ZNF193), homeobox genes (PHOX2B, CDX1), Fig. 3 Normalized microarray expression values for MMSET in t(4; and forkhead box genes (FOXO3A). 14) (black bars) and non-t(4;14) myeloma patients. Five patients with- Apoptosis and Related Genes. A number of genes im- out a t(4;14) express MMSET to levels similar to those of patients with portant in protecting the cell from apoptosis were also identified this translocation.

Fig. 4 The 24 exons of MMSET, the position of the IgH-MMSET primer and Af- fymetrix probe set are shown diagrammatically. Shaded boxes, the position of the initiation sites; the three possible protein isoforms are also shown (types I, II, and III).

(leukosialin, sialophorin), which has a prominent role in cell– DNA between nucleosomes and functions in the compaction of cell adhesion. We found three differentially expressed chemo- chromatin into higher order structures. ␣-Thalassemia/mental kine-related genes: CCL18, CCRL2, and CCL13, which were all retardation syndrome X-linked (RAD54) homolog belongs to down-regulated in the cases with a t(4;14), suggesting that there SWI/SNF family of chromatin remodeling proteins and is in- may be differences in distribution of the myeloma plasma cells volved in DNA methylation and gene regulation of interphase between the two subsets (t(4;14) versus non-t(4;14)). Differen- and chromosomal segregation in mitosis. tial patterns of cytokine response could mediate differential DNA Repair Gene Differences. We found only one disease activity and response to therapy; however, we only DNA repair gene that was differentially expressed between noted altered levels of two cytokines receptors, IL9R and FGFR3-positive and -negative cases. APEX nuclease (apurinic/ IL10RA. Both receptors were noted to be down-regulated in apyrimidinic endonuclease) 2 (APEX2) is a Class II AP endo- t(4;14) cases. Several studies have reported that interleukin nuclease that cleaves the phosphodiester DNA backbone 5Ј to (IL)-10 affects myeloma cells by stimulating secondary signals AP sites (sites that occur after DNA damage) and that provides for cell proliferation through oncostatin M (OSM) and IL-11, a scaffold support for the activity of a number of subsequent whereas IL-9 has not, thus far, been implicated in the disease repair processes. process. The ATP-binding cassette transporter genes are important in mediating membrane transport of a wide range of sub- stances, including drugs across the cell membrane. ABCB1 DISCUSSION (previously known as P-glycoprotein/multidrug resistance 1) is Gene array studies performed previously in myeloma have well recognized for its ability to mediate resistance to a range of demonstrated that malignant plasma cells can be distinguished chemotherapeutic drugs. We found this to be underexpressed in from normal plasma cells by their gene expression profiles, and t(4;14) plasma cells, which suggests that any drug resistance that CD138-selected plasma cells from monoclonal gammopa- occurring in these patients is mediated via another mechanism. thy of uncertain significance are more similar to their malignant Oncogenes. We found altered expression of a number of counterparts than to their normal ones (14, 20–22). Further- oncogenes, including Wilms tumor 1 and B-cell CLL/lymphoma more, hierarchical clustering of myeloma cases confirms that 7B (BCL7B), which are both down-regulated, and v-myb myelo- myeloma is a heterogeneous disease at the molecular level and blastosis viral oncogene homolog (avian)-like 1 (MYBL1), that distinct groups of cases can be defined on the basis of the which is up-regulated. patterns of gene expression (23, 24). In the current analysis, we Chromatin. Two genes involved in chromatin structure clearly show that cases carrying a t(4;14) can be defined as acetylation/methylation of DNA were differentially expressed being different from cases lacking the translocation. The gene between the two groups. Histone 1 H3d,(H1STAH3d)isin- signature typical of the t(4;14) cases is limited and equates to volved with nucleosome assembly by interacting with linker 127 genes, 67 of which are up-regulated, whereas the remainder

Table 2 Genes with significant mean differences between t(4;14) and non-t(4;14) cases arranged according to function (functions derived from LocusLink http://www.ncbi.nlm.nih.gov/LocusLink) Mean Gene Name Accession no. Functions difference P value Membrane-associated genes Chemokine-related genes CCRL2 Chemokine (C-C motif) receptor-like 2 AF014958 Transmembrane protein related to CCR1 Ϫ1.03 0.001 CCL13 Chemokine (C-C motif) ligand 13 AJ001634 Accumulation of leukocytes during Ϫ1.07 0.005 inflammation CCL18 Chemokine (C-C motif) ligand 18 Y13710 Attracts naïve T-cells to dendritic cells and Ϫ1.11 0.017 (pulmonary and activation-regulated) activated macrophages in lymph nodes FY Duffy blood group X85785 Facilitates movement of chemokines across Ϫ1.08 0.002 endothelium Cytokine related genes AREG Amphiregulin (schwannoma-derived growth M30704 EGF family protein, cell-signaling, growth 1.45 0.036 factor) factor activity IL10RA Interleukin 10 receptor, ␣ U00672 Survival of progenitor myeloid cells Ϫ1.08 0.003 through the insulin receptor substrate-2/ PI 3-kinase/AKT pathway IL9R Interleukin 9 receptor L39064 JAK/STAT activation after ligand binding Ϫ1.04 0.001 LILRB5 Leukocyte immunoglobulin-like receptor, AF025534 Immunoglobulin superfamily member Ϫ1.13 0.016 subfamily B (with TM and ITIM domains), member 5 Transporters SLC9A2 Solute carrier family 9 (sodium/hydrogen S81591 Ion transport, pH regulation Ϫ1.07 0.010 exchanger), isoform 2 ABCB1 ATP-binding cassette, sub-family B (MDR/ M14758 Encodes P-glycoprotein drug efflux pump Ϫ1.33 0.002 TAP), member 1 TMP21 Transmembrane trafficking protein L40397 Protein carrier integral to plasma membrane, 1.25 0.024 involved in ER to Golgi transport L40391 1.19 0.005 SCN9A Sodium channel, voltage-gated, type IX, ␣ X82835 Sodium ion, and possible calcium ion, 1.21 0.048 polypeptide transporter FOLR1 Folate receptor 1 (adult) U20391 GPI-anchored plasma membrane-bound Ϫ1.13 0.002 receptor, folic acid endocytosis DJ971N18 Hypothetical protein DJ971N18 AL021396 Membrane electron transport 1.19 0.029 P5 Protein disulfide isomerase-related protein D49489 Substrate-specific chaperone activity 1.17 0.001 FTHP1 Ferritin, heavy polypeptide 1 J04755 Intracellular iron ion transport and storage, 1.14 0.011 part of the ferritin complex ATP4A ATPase, Hϩ/Kϩ exchanging, ␣ M63962 Small molecule transport, ATPase activity Ϫ1.01 0.001 polypeptide FADS1 Fatty acid desaturase 1 AC004770 Possible electron transport Ϫ1.1 0.002 Adhesion SPN Sialophorin (gpl 115, leukosialin, CD43) J04168 Negative regulator of cell adhesion, 1.42 0.017 chemotaxis, cell polarity, possible T- cell activation ALCAM Activated leukocyte cell adhesion molecule Y10183 Signal transduction, down regulated in high 1.24 0.019 grade prostate tumors, may have a role in progression CDH2 Cadherin 2, type 1, N-cadherin (neuronal) M34064 Cell–cell adhesion, calcium dependent, 1.17 0.040 plasma membrane glycoprotein CTNND2 Catenin (cadherin-associated protein), ␦ 2 U96136 Neuronal cell adhesion, deletion associated Ϫ1.01 0.003 (neural plakophilin-related arm-repeat with cri-du-chat syndrome protein) CSPG2 Chondroitin sulphate proteoglycan 2 X15998 Binds hyaluronic acid, heterophilic cell Ϫ1.04 0.010 (versican) adhesion, melanoma progression CD99 CD99 antigen M16279 Engagement triggers exocytosis of Ϫ1.04 0.004 ganglioside GM1 and reorganization of actin cytoskeleton DCN Decorin M14219 Small matrix proteoglycan, connective 1.3 0.002 tissue component, binds type I collagen VIM Vimentin Z19554 Linked to motility, metastasis and invasive 1.27 0.010 phenotype, promoter is a target of the ␤-catenin/TCF pathway Other membrane genes GAGE5 G antigen 5 U19146 Cancer testis antigen Ϫ1.02 0.003

(Table continues)

Table 2 Continued. Mean Gene Name Accession no. Functions difference P value Signaling: Growth Ras/MAPK RASGRP1 Ras guanyl releasing protein 1 (calcium AF081195 Signal transduction, calcium ion binding, 1.42 0.028 and DAG-regulated) lipid binding, Ras activation ARHQ ras homolog gene family, member Q M31470 Interacts with exocyst complex components 1.34 0.024 ARHH ras homolog gene family, member H Z35227 Small G-like protein expressed in 1.19 0.04 hematopoietic cells and tissues RPIB9 Rap2 binding protein 9 AI825798 Novel gene Ϫ1.06 0.007 MAPK14 mitogen-activated protein kinase 14 L35263 Stress response, cell surface receptor linked Ϫ1.19 0.002 signal transduction SH3BGRL SH3 domain binding glutamic acid-rich AF042081 Contains SH3 binding motif 1.28 0.031 protein like Protein tyrosine phosphatases DUSP10 dual specificity phosphatase 10 AB026436 MAPK superfamily, binds and inactivates 1.24 0.004 p38 and SAPK/JNK PTP4A1 Protein tyrosine phosphatase type IVA, U48296 Prenylated protein, over expression confers 1.24 0.014 member 1 transformed phenotype PPP2R1A Protein phosphatase 2 (formerly 2A), J02902 Ser/Thr phosphatase, implicated in negative Ϫ1.04 0.001 regulatory subunit A (PR65), ␣ isoform control of growth and division PTN Pleiotrophin (heparin binding growth M57399 Has protein phosphatase inhibitor activity, Ϫ1.29 0.001 factor 8, neurite growth-promoting over expression associated with factor 1) inflammation ϩ pancreatic cancer Cell cycle CCND2 Cyclin D2 D13639 Regulatory subunit of CDK4 or CDK6, 1.17 0.002 G1/S transition, Rb phosphorylation, highly expressed in ovarian and testicular tumors UBE2C Ubiquitin-conjugating enzyme E2C U73379 Required for destruction of mitotic cyclins Ϫ1.07 0.001 and cell cycle progression RBMS1 RNA binding motif, single stranded X77494 Protein family implicated in DNA 1.4 0.002 interacting protein 1 replication, transcription, cell cycle progression, apoptosis CALM1 Calmodulin 1 (phosphorylase kinase, ␦) U16850 Phospholipase C-␤ interacting protein, 1.16 0.020 functions in growth and cell cycle, and signal transduction BS69 Adenovirus 5 E1A binding protein AA127624 Transcriptional repressor, negative 1.34 0.019 regulation of cell cycle FGF-related FRAG1 FGF receptor activating protein 1 AL049261 Predicted tumor suppressor, integral to Ϫ1 0.005 membrane GARP Glycoprotein A repetitions predominant Z24680 Type 1 membrane protein, gene near to Ϫ1.03 0.001 FGF and TRY loci AREG amphiregulin (schwannoma-derived growth M30704 EGF family protein, autocrine growth 1.45 0.036 factor) factor for fibroblasts PRKCB1 Protein kinase C, ␤ 1 X07109 Amino acid phosphorylation 1.17 0.038 Signaling: Survival Apoptosis CASP2 Caspase 2, apoptosis-related cysteine U13021 Apoptotic program, activation induced by Ϫ1.22 0.001 protease variety of apoptotic stimuli DEDD Death effector domain containing AF043733 DEDs are shared by adaptors, regulators Ϫ1.01 0.001 and executors of apoptosis pathway, overexpression can induce weak apoptosis FOXO3A Forkhead box O3A AF032886 Transcription factor for genes necessary for 1.18 0.008 cell death HDAC1 Histone deacetylase 1 U50079 Interacts with retinoblastoma protein; cell Ϫ1.07 0.004 proliferation, and differentiation control; deacetylates and modulates the effects of p53 associated with MTA2 IER3 Immediate early response 3 S81914 Protection against Fas- or TNF␣-induced Ϫ1.16 0.001 apoptosis Stress DUSP10 Dual specificity phosphatase 10 AB026436 Negative regulator of MAPK family 1.24 0.004 members MAPK14 Mitogen-activated protein kinase 14 L35263 Activated by stress, stress related Ϫ1.19 0.002 transcription and genotoxic stress response CASP2 Caspase 2, apoptosis-related cysteine U13021 Regulator of apoptosis, can be activated by Ϫ1.22 0.001 protease cytotoxic stress

Table 2 Continued. Mean Gene Name Accession no. Functions difference P value Nuclear Oncogenes WHSC1 Wolf-Hirschorn syndrome candidate 1 AJ007042 May be transforming event of t(4;14) 1.27 0.005 (MMSET) PTP4A1 Protein tyrosine phosphatase type IVA, U48296 Overexpression confers transformed 1.24 0.014 member 1 phenotype ARHH Ras homolog gene family, member H Z35227 Small G-like protein expressed in 1.19 0.04 hematopoietic cells and tissues, oncogenesis Tumor suppressors FRAG1 FGF receptor activating protein 1 AL049261 Predicted tumor suppressor Ϫ1 0.005 Chromatin ATRX ␣ thalassemia/mental retardation syndrome U72936 Chromatin remodeling, DNA methylation, 1.15 0.039 X-linked (RAD54 homolog) recombination and repair 1.16 0.030 HIST1H3D Histone 1, H3d N35832 Nucleosome assembly Ϫ1.06 0.032 Transcription CSDA Cold shock domain protein A M24069 Negative regulation of transcription from 1.57 0.014 Pol II promoter MATR3 Matrin 3 AB018266 May play a support structure role for 1.36 0.020 transcription BS69 Adenovirus 5 E1A binding protein AA127624 Negative regulation of transcription from 1.34 0.019 Pol II promoter MITF Micropthalmia-associated transcription AB006909 Transcriptional activator activity 1.23 0.010 factor NRIP1 Nuclear receptor interacting protein 1 X84373 Modulates transcriptional activity of the 1.21 0.029 ER, co-activator activity FOXO3A Forkhead box O3A AF032886 transcription factor for genes necessary for 1.18 0.008 cell death MYBL1 v-myb myeloblastosis viral oncogene X66087 Transcriptional activator 1.12 0.030 homolog (avian)-like 1 WT1 Wilms tumor 1 X51630 Transcription factor, functions in kidney Ϫ1.01 0.001 and gonad proliferation and differentiation PHOX2B Paired-like homeobox 2b D82344 Transcription factor, development of some Ϫ1.07 0.003 neuron populations RNPC2 RNA-binding region (RNP1, RRM) L10910 RNA binding protein and possible splicing 1.1 0.012 containing 2 factor DEDD Death effector domain containing AF043733 Negative regulation of transcription Ϫ1.01 0.001 TIEG2 TGFB inducible early growth response 2 AI991531 Negative regulation of transcription from Ϫ1.03 0.002 Pol II promoter CDX1 Caudal type homeobox transcription U51095 Transcription factor activity, down Ϫ1.04 0.001 factor 1 regulated after promoter methylation in colorectal tumor cell lines ZFP36L1 Zinc finger protein 36, CH3 type-like 1 X79067 Putative nuclear transcription factor 1.44 0.01 ZNF193 Zinc finger protein 193 AL022393 Transcription factor activity Ϫ1 0.008 DNA repair ATRX ␣ thalassemia/mental retardation syndrome U72936 Chromatin remodeling, DNA methylation, 1.15 0.039 X-linked (RAD54 homolog) recombination and repair 1.16 0.030 APEX2 APEX nuclease (apurinic/apyrimidinic AJ011311 Nuclear and mitochondrial base excision Ϫ1.08 0.007 endonuclease)2 repair, protein has weak class II AP endonuclease activity Abbreviations: EGF, epidermal growth factor; TM, trans membrane; ITIM, immunoreceptor tyrosine-based inhibition motif ; ER, estrogen receptor; TCF, T cell factor; MAPK, mitogen-activated protein kinase; DED, death effector domain; MTA2, metastasis-associated gene family member 2; RRM, RNA recognition motif; TGFB, transforming growth factor beta. are down-regulated, and contains genes that may explain differ- The t(4;14) in multiple myeloma results in a reciprocal ences in the biology and outcome of patients with the translo- translocation between chromosome 14q32 and 4p16.3, reposi- cation. Two of the genes highlighted in the supervised analysis tioning FGFR3 to der(14) and creating a fusion gene with were also identified in an unsupervised analysis. The unsuper- MMSET on der(4) under the influence of strong enhancers from vised analysis used a list comprising genes that varied most the immunoglobulin heavy chain gene region. FGFR3 is one of across the sample set. The comparison of these two analyses a family of five tyrosine kinases through which the fibroblast illustrates the fact that there is a core set of genes which are growth factors signal. These receptors are characterized by an deregulated by the translocation and contribute to the overall extracellular domain with either two or three immunoglobulin- heterogeneity of myeloma. like domains, a transmembrane domain, and a cytoplasmic

tyrosine kinase domain. On ligand stimulation FGFR3 under- eloma. Constitutive expression of FGFR3 would be predicted to goes dimerization and tyrosine autophosphorylation, resulting in signal via the RAS/MAPK and JAK/STAT pathway. Genes proliferation and differentiation, depending on the cellular con- involved in these pathways were noted to be differentially text. FGFR3 is dysregulated by mutations in a number of tumor expressed between cases with and without a translocation. Re- systems and in inherited conditions associated with abnormal cent studies have suggested that cyclin expression is important bone growth and development. These mutations are associated in myeloma pathogenesis, with different myeloma translocation with constitutive activation of the tyrosine kinase activity of the subgroups being characterized by their use of cyclin D1, cyclin gene; however, in myeloma patients such mutations are ex- D2,orcyclin D3, and the expression of cyclin D2 differentiating tremely rare and deregulation seems to occur exclusively as a our two subgroups would be in keeping with these data (18). consequence of the translocation (11, 25). The possible onco- Glycoprotein A repetitions predominant (GARP), a FGF-related genic nature of FGFR3 in myeloma has been examined by a gene, was noted to be down-regulated in the t(4;14) cases. It is number of groups (26–28). These studies demonstrate that located in close proximity to cyclin D1 at 11q13–11q14, a gene FGFR3 is capable of producing lymphoid malignant disease in frequently deregulated by the t(11;14) translocation. In the un- mice and that overexpression in myeloma cell lines results in supervised analysis, overexpression of cyclin D1 and cyclin D2 increased cell proliferation and survival. Recent studies also were demonstrated to be mutually exclusive. Data from cyclin suggest that the inhibition of FGFR3 by a small molecule D1 transgenic mice and transfection studies demonstrate that induces differentiation and apoptosis of t(4;14) myeloma cell overexpression of cyclin D1 is insufficient to cause cell cycle lines, suggesting that FGFR3 may be a possible therapeutic progression alone, and it has been suggested that overexpression target (29). of cyclin D1 renders a cell more sensitive to growth-activating In contrast to FGFR3, the role of MMSET in myeloma signals and/or less sensitive to growth-inhibitory signals (30, pathogenesis has been more difficult to understand. MMSET is 31). In this study, although a number of growth-activating and a large gene of 90 kb that comprises 25 exons, is expressed growth-inhibitory genes were highlighted as being altered be- ubiquitously in early development, and undergoes complex al- tween cases with and without a translocation, from the literature, ternative splicing. The gene encodes a 136 kDa protein contain- no genes appeared to have a direct link with cyclin pathways. A ing four domains: a PWWP domain, an HMG box, a SET number of genes important in protecting the cell from apoptosis domain and a PHD-type zinc finger. A literature and database were identified as being differentially expressed between cases search identifies nine possible cDNA variants by using the with and without a t(4;14) translocation. This suggests that, in standard criteria of ATG and TATA box on the 5Ј side to addition to the known overexpression of cyclins, which drives identify initiation sites. Two open reading frames are identifia- entry into the cell cycle, disturbances of pathways associated ble: the first in exon 3, translating to produce a full-length with apoptosis contribute to the development of B-cell malig- isoform of protein (1,365 aa, type II), and four possible trun- nant diseases. Of particular interest, death effector domain- cated variants (647–802 aa, type I), all containing the COOH- containing gene (DEDD) was noted to be down-regulated in terminal hath and HMG domains. A second open reading frame cases with the translocation. The death effector domain (DED) in exon 15 can also produce a hypothetical protein product of is a protein–protein interaction domain shared by adaptors,

584 aa containing the NH2-terminal hath and SET domains regulators, and executors of the programmed cell death path- (type III or ILS promoter REII region-binding protein). way, and overexpression can induce apoptosis. DED occurs in We have previously described the sites of the breakpoints proteins that regulate programmed cell death, and DED inter- on chromosome 4 and can relate these breakpoints to the po- actions connect with the caspases to mediate formation of the tential isoforms of MMSET which can be expressed. The probes death-inducing signal complex. Accumulating evidence now on the gene chip are located in exon 24, and all cases with a suggests that DED-containing proteins have additional roles in t(4;14) are positive for this area. The primers for the IgH- controlling pathways of cellular activation and proliferation (32). MMSET PCR are directed to joining gene segment 6 (JH6). In Non-homologous end joining (NHEJ) DNA double-strand this case series, IgH is joined to MMSET exon 3 in two patients, break (DSB) repair pathways are integral components of class exon 4 in another two patients, and exon 5 in one patient. This switch recombination, and we postulated that there may be suggests that all of the patients are capable of expressing the 3Ј abnormalities in these pathways in myeloma cases with aberrant sequence and, depending on where the breakpoint is situated class switching and translocations involving chromosome 14. with reference to the initiation site in exon 3, may express the 5Ј Previously, we have reported that two components of the NHEJ sequence. This would result in some patients with a t(4;14) pathway, XRCC4 and RAD50, are significantly up-regulated in producing the type I, II, and/or III protein, whereas other pa- myeloma cases compared with normal controls, suggesting a tients could only produce the type III protein. Twenty-five role for these DNA repair enzymes in the etiology of myeloma. percent of cases that did not have the IgH-MMSET fusion In this study, we found only one DNA repair gene that was transcript also expressed the 3Ј end of the MMSET transcript. differentially expressed between FGFR3-positive and -negative Further analysis of these cases is required, but in this initial cases, APEX2. This is in keeping with our previous data where study, the results suggest that these cases express normal we used a quantitative TaqMan PCR approach to explore the MMSET (type II) or the type III isoform (IREII-BP) only. patterns of DSB repair genes in a larger series of cases (33). This The downstream consequences of FGFR3 up-regulation or study included a number of important members of the NHEJ IgH-MMSET fusion transcripts is unknown; thus, characterizing pathway that are not present on the Affymetrix U95Av2 chip. the differentially expressed genes may help to understand the Expression of these genes was similar between cases with and consequences of abnormal signaling via these pathways in my- without a t(4;14) translocation, suggesting that there are distinct

patterns of gene expression that distinguish myeloma plasma 14. Davies FE, Dring AM, Li C, et al. Insights into the multistep cells from other lymphoid malignant disease, but the differences transformation of MGUS to myeloma using microarray expression between cases of myeloma are limited. analysis. Blood 2003;102:4504–11. In conclusion, using RT-PCR and gene analysis, we dem- 15. Li C, Wong WH. Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc Natl Acad Sci onstrate that cases with a t(4;14) have a distinct gene expression 2001;98:31–6. pattern compared with other cases of myeloma. Importantly a 16. Li C, Wong WH. Model-based analysis of oligonucleotide arrays: number of pathways are dysregulated as a consequence of the model validation, design issues and standard error application. Genome t(4;14) in multiple myeloma, and two genes, WHSC1/MMSET Biol 2001;2:research0032. and cyclin D2, appear to play a central role and may account for 17. Tian E, Zhan F, Walker R, et al. The role of the Wnt-signaling the poor prognosis associated with this translocation. Interest- antagonist DKK1 in the development of osteolytic lesions in multiple ingly, 25% of myeloma cases lacking evidence of a transloca- myeloma. N Engl J Med 2003;349:2483–94. tion by RT-PCR also had up-regulation of a MMSET transcript 18. Bergsagel PL, Kuehl WM. Critical roles for immunoglobulin trans- to the same level as cases with a translocation, suggesting that locations and cyclin D dysregulation in multiple myeloma. Immunol Rev 2003;194:96–104. this gene plays an important role in myeloma pathogenesis. 19. Hofmann WK, de Vos S, Tsukasaki K, Wachsman W, Pinkus GS, Said JW. Altered apoptosis pathways in mantle cell lymphoma detected REFERENCES by oligonucleotide microarray. Blood 2001;98:787–94. 1. Avet-Loiseau H, Facon T, Grosbois B, et al. Oncogenesis of multiple 20. Claudio JO, Masih-Khan E, Tang H, et al. A molecular compen- myeloma: 14q32 and 13q chromosomal abnormalities are not randomly dium of genes expressed in multiple myeloma. Blood 2002;100:2175–86. distributed, but correlate with natural history, immunological features, 21. De Vos J, Thykjaer T, Tarte K, et al. Comparison of gene expres- and clinical presentation. Blood 2002;99:2185–91. sion profiling between malignant and normal plasma cells with oligo- 2. Fonseca R, Debes-Marun CS, Picken EB, et al. The recurrent IgH nucleotide arrays. Oncogene 2002;21:6848–57. translocations are highly associated with nonhyperdiploid variant mul- 22. Zhan F, Hardin J, Kordsmeier B, et al. Global gene expression tiple myeloma. Blood 2003;102:2562–7. profiling of multiple myeloma, monoclonal gammopathy of undeter- 3. Keats JJ, Reiman T, Maxwell CA, et al. In multiple myeloma, mined significance, and normal bone marrow plasma cells. Blood 2002; t(4;14)(p16;q32) is an adverse prognostic factor irrespective of FGFR3 99:1745–57. expression. Blood 2003;101:1520–9. 23. Magrangeas F, Nasser V, Avet-Loiseau H, et al. Gene expression 4. Rasmussen T, Hudlebusch HR, Knudsen LM, Johnsen HE. FGFR3 profiling of multiple myeloma reveals molecular portraits in relation to dysregulation in multiple myeloma: frequency and prognostic relevance. the pathogenesis of the disease. Blood 2003;101:4998–5006. Br J Haematol 2002;117:626–8. 24. Shaughnessy JD Jr., Barlogie B. Interpreting the molecular biology 5. Chesi M, Nardini E, Brents LA, et al. Frequent translocation t(4; and clinical behavior of multiple myeloma in the context of global gene 14)(p16.3;q32.3) in multiple myeloma is associated with increased expression profiling. Immunol Rev 2003;194:140–63. expression and activating mutations of fibroblast growth factor receptor 3. Nat Genet 1997;16:260–4. 25. Intini D, Baldini L, Fabris S, et al. Analysis of FGFR3 gene mutations in multiple myeloma patients with t(4;14). Br J Haematol 6. Richelda R, Ronchetti D, Baldini L, et al. A novel chromosomal 2001;114:362–4. translocation t(4; 14)(p16.3; q32) in multiple myeloma involves the fibroblast growth-factor receptor 3 gene. Blood 1997;90:4062–70. 26. Chesi M, Brents LA, Ely SA, et al. Activated fibroblast growth factor receptor 3 is an oncogene that contributes to tumor progression in 7. Chesi M, Nardini E, Lim RS, Smith KD, Kuehl WM, Bergsagel PL. multiple myeloma. Blood 2001;97:729–36. The t(4;14) translocation in myeloma dysregulates both FGFR3 and a novel gene, MMSET, resulting in IgH/MMSET hybrid transcripts. 27. Li Z, Zhu YX, Plowright EE, et al. The myeloma-associated onco- Blood 1998;92:3025–34. gene fibroblast growth factor receptor 3 is transforming in hematopoi- 8. Bergsagel PL, Kuehl WM. Chromosome translocations in multiple etic cells. Blood 2001;97:2413–9. myeloma. Oncogene 2001;20:5611–22. 28. Plowright EE, Li Z, Bergsagel PL, et al. Ectopic expression of 9. Sibley K, Fenton JA, Dring AM, Ashcroft AJ, Rawstron AC, Morgan fibroblast growth factor receptor 3 promotes myeloma cell proliferation GJ. A molecular study of the t(4;14) in multiple myeloma. Br J Haema- and prevents apoptosis. Blood 2000;95:992–8. tol 2002;118:514–20. 29. Trudel S, Ely S, Farooqi Y, et al. Inhibition of fibroblast growth 10. Fenton JAL, Pratt G, Rawstron AC, et al. Genomic characterization factor receptor 3 induces differentiation and apoptosis in t(4;14) my- of the chromosomal breakpoints of t(4;14) of multiple myeloma sug- eloma. Blood 2004;103:352–8. [Epub ahead of print, 2004 Jan 8]. gests more than one possible aetiological mechanism. Oncogene 2003; 30. Bodrug SE, Warner BJ, Bath ML, Lindeman GJ, Harris AW, 22:1103–13. Adams JM. Cyclin D1 transgene impedes lymphocyte maturation and 11. Onwuazor ON, Wen XY, Wang DY, et al. Mutation, SNP, and collaborates in lymphomagenesis with the myc gene. EMBO J 1994; isoform analysis of fibroblast growth factor receptor 3 (FGFR3) in 150 13:2124–30. newly diagnosed multiple myeloma patients. Blood 2003;102:772–3. 31. Lovec H, Grzeschiczek A, Kowalski MB, Moroy T. Cyclin D1/ 12. Santra M, Zhan F, Tian E, Barlogie B, Shaughnessy J Jr. A subset bcl-1 cooperates with myc genes in the generation of B-cell lymphoma of multiple myeloma harboring the t(4;14)(p16;q32) translocation lacks in transgenic mice. EMBO J 1994;13:3487–95. FGFR3 expression but maintains an IGH/MMSET fusion transcript. 32. Barnhart BC, Lee JC, Alappat EC, Peter ME. The death effector Blood 2003;101:2374–6. domain protein family. Oncogene 2003;22:8634–44. 13. Malgeri U, Baldini L, Perfetti V, et al. Detection of t(4;14)(p16.3; 33. Roddam PL, Allan JA, Davies FE, Worrillow LJ, Dring AM, q32) chromosomal translocation in multiple myeloma by reverse tran- Morgan GJ. Non-homologous end joining pathway is differentially scription-polymerase chain reaction analysis of IGH-MMSET fusion regulated in malignant plasma cells compared to other lymphoid tu- transcripts. Cancer Res 2000;60:4058–61. mours. Blood 2003;102:2138 Part 1.

Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2004 American Association for Cancer Research. A Global Expression-based Analysis of the Consequences of the t(4;14) Translocation in Myeloma

Ann M. Dring, Faith E. Davies, James A. L. Fenton, et al.

Clin Cancer Res 2004;10:5692-5701.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/17/5692

Cited articles This article cites 32 articles, 18 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/17/5692.full#ref-list-1

Citing articles This article has been cited by 10 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/17/5692.full#related-urls

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/10/17/5692. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.