ALBERT

Description: Abstract 773 Peripheral T-cell lymphomas (PTCLs) are rare and heterogeneous tumors whose biology is largely unknown. Interestingly, the commonest subtypes (i.e. PTCL not otherwise specified, NOS; angioimmunoblastic T-cell lymphoma, AITL; and anaplastic large cell lymphoma, ALCL) present on one hand few disease-specific molecular features and, on the other hand, several apparently common abnormalities. So far, no data are available regarding miRNA expression in these tumors. In order to identify miRNA deregulated in PTCLs, we performed an extensive miRNA profiling (by studying 379 targets on the TaqMan Array MicroRNA Cards) of 44 PTCLs (including 23 PTCLs/NOS, 12 ALCLs, and 9 AITLs) and 13 sample representative of normal T-cell sub-populations (CD4+ and CD8+, both resting and activated). In addition, for all these cases, gene expression profiles (GEPs) were generated by the Ilumina whole genome DASL-assay. TaqMan Quantitative-PCR (qPCR) was then used for validation. First, we found that PTCLs and normal T-cells could be easily distinguished based on their miRNA profile, by both unsupervised and supervised analysis. Specifically, the latter identified 91 miRNA differentially expressed in PTCLs vs. T-cells with a fold change ≥2 and a pvalue

Description: Abstract 2494 Background. Burkitt lymphoma (BL) is currently listed in the WHO classification of lymphoid tumors as a single genetic and morphological entity with variation in clinical presentation. In particular, three clinical subsets of BL are recognized: endemic (eBL), sporadic (sBL) and immunodeficiency associated (ID-BL). Each affects different populations and can present with different features. So far, possible differences in their gene expression profiles (GEP) have not been investigated. In this study we aimed to 1) assess whether BL subtypes present with differences in their GEP; 2) investigate the relationship of the different BL subtypes with the non-neoplastic cellular counterparts; 3) Identify genes and programs specifically deregulated in BLs and possibly contributing to the malignant phenotype. Methods. We studied by GEP 128 cases of B-cell derived malignancies and 20 samples of normal B-cell subpopulations GEP analysis. In particular, we included 40 BLs (13 eBLs, 21 sBLs 6 HIV-BLs), 40 follicular lymphomas, 10 chronic lymphocytic leukemias, 10 GCB-type diffuse large B-cell lymphomas, 10 ABC-type DLBCL, 5 primary mediastinal B-cell lymphomas, 13 HIV-related DLBCL, as well as 10 germinal center (GC), 5 naïve and 5 memory cells samples. GEP results were confirmed by dividing BL cases into training and test subgroups. In addition, as further validation, we performed immunohistochemistry (IHC) on tissue microarrays containing 85 BL cases as well as functional assays in vitro and in vivo, by focusing on the role of RBL2, a tumor suppressor gene involved in cell cycle control and mutated in eBL. Specifically, we used cell transfection and shRNAs (for mimicking MYC over-expression and RBL2 silencing), soft agar and invasion capability assays, and xenografted mouse models. Results. First, we found that BLs constitute a unique molecular entity, with a relatively homogeneous GEP, distinct from other B-cell malignancies. Indeed, by unsupervised analysis all BLs clearly clustered apart of other lymphomas. However, by supervised analysis, we found that BL subtypes presented slight differences in their GEPs. Particularly, eBLs and ID-BLs appeared to be almost identical, diverging from sBLs. Specifically, they varied for genes involved in cell cycle control, BCR-signaling, and TNF/NFKB-pathways. Of note, eBLs and ID-BLs on one hand, and sBLs on the other (roughly corresponding to EBV+ vs. EBV− cases) also differed for genes target of mi-R127a, which is altered in EBV+ cases as a direct consequence of viral integration. To further investigate cell cycle regulation in BLs, we inferred a network of RBL2-depending genes by reverse engineering, by uncovering possible RBL2 transcriptional targets. Interestingly, we found that eBL and sBL diverged for genes belonging to such network. Notably, we provided evidences that RBL2 can cooperate with MYC in inducing a neoplastic phenotype in vitro and in vivo. In particular, lymphoblastoid cells engineered to carry both MYC over-expression and RBL2 silencing presented with increased colony formation and matrix invasion capabilities, and higher efficiency in inducing tumor formation in nude mice if compared to single transfectants (MYC+ or RBL2−). Moreover, as the present WHO classification does not definitely identify the counterpart of eBL, we compared BLs GEP to those of normal B-cells. We found that all BL subtypes were intimately related to GC cells (by showing an early stage GC differentiation arrest), differing from them for molecules specially involved in cell proliferation, immune response, and signal transduction. Finally, as further validation of GEP, we studied by IHC the expression of SPARC and CYR61, two molecules involved in human tumorigenesis. Indeed, they turned out to be consistently expressed by neoplastic elements in all instances, as indicated by GEP analysis. Conclusions. Our study provided substantial insights on the pathobiology of BLs, by offering novel evidences which may be relevant for its classification and possibly future treatment. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 1304 Recently, a class of noncoding RNAs called microRNA (miRNAs) has emerged as critical gene regulators in cell growth, differentiation, disease and development. MiRNAs are 18–24 nucleotide long noncoding RNAs, which regulate gene expression by pairing with 3′ untranslated region (UTR) of target mRNA and inhibiting protein translation and/or inducing mRNA degradation. Deregulated miRNA expression is reported in various human diseases including lymphomas, suggesting an important role in their pathogenesis. According to WHO classification, Burkitt lymphoma (BL) is a rare, highly aggressive NHL composed of monomorphic medium-sized B cells with multiple nucleoli and numerous mitotic figures and is more common in children than in adults. The molecular feature of BL is the translocation that places MYC under the control of immunoglobulin gene regulatory elements. High levels of c-MYC have been clearly shown to have a tumour-promoting effect. However, there is recent evidence that infrequent cases may lack an identifiable MYC translocation, the explanation for which is still uncertain, though suggesting the existence of pathogenetic mechanisms alternative to genetic alterations. Over the past years miRNA signatures have been described to characterize and classify different types of BL or to investigate the expression of miRNAs possibly regulated by c-Myc in BL cases positive or negative for Myc translocation. However, it remained unclear the functional role of differentially expressed miRNAs and no further studies have been conducted. We performed miRNA expression profile to gain further insights into the molecular pathology of BL. We conducted array analysis on a set of 5 sporadic BL patients, 3 endemic BL patients, 9 reactive tissues and 11 cases of mononucleosis. Our profile is the first one that shows the different expression between BL cases and normal B cells whereas recent miRNA profiles have been conducted in BL compared to other B-NHL (B-CLL, MCL & FL). A common trend of miRNAs altered expression was also observed by NanoString analysis in 10 BL cell lines compared to 5 normal CD-19+ B cells. Among several miRNAs previously described be deregulated in BL we identified a severe down-regulation of miR-221, miR-222 in all classes of comparisons we analyzed. The down-regulation of miR-221 and miR-222 associated to BL has been also confirmed by q-RT-PCR method in a different cohort of BL patients (20) compared to the healthy controls (6). We found that interesting considering the up-regulation of miR-221 and miR-222 previously confirmed in a lot of solid tumors by multiple studies. We are investigating a different role of the cluster miR-222 and miR-221 in lymphomas that have a different process in carcinogenesis than solid tumors. In vivo models to study the lymphomagenesis of BL have been created but until now no one studied the importance of the miRNAs in vivo. We analyzed the expression of miR-221 and miR-222 in a Myc transgenic mouse model. The transgene construct consists of the Myc oncogene (c-myc) in association with the Emu immunoglobulin heavy chain enhancer and Myc promoter. Expression of the mouse Myc transgene is restricted to the B cell lineage. Previously it has been shown an increase of pre-B cells in the bone marrow throughout life of hemizygotes and a transient increase in large pre-B cells in the blood at 3–4 weeks of age; moreover spontaneous pre-B and B cell lymphomas reach an incidence of 50% at 15–20 weeks in hemizygous progeny of a wildtype female mated with a hemizygous male. We observed the development of Burkitt lymphoma within 10 weeks of birth in 14 out of 25 Eu-Myc transgenic mice and a premature death in 5 out or 25 transgenic mice within 6–8 weeks of birth without showing any enlarged lymph nodes. Transgenic mice with masses showed the same phenotype characterized by enlarged spleen (3 fold), lymphosarcomas associated with BL and enlarged lymph nodes around the neck area. B-cells have been negatively selected from enlarged lymph nodes and enlarged spleen. A qRT-PCR has been conducted to evaluate the miR-221 and miR-222 expression. The miRNA levels showed a down-regulation in B cells collected from the masses when compared to normal B cells derived from the spleen of WT mice. In conclusion, our study reveals new insights into the functional significance in loss of miR-221 and miR-222 expression in BL pathogenesis. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 4631 Background Peripheral T cell lymphomas (PTCLs) are a heterogeneous group of tumors representing approximately 12% of lymphoid neoplasms, basically subdivided into specified and not specified (NOS) forms. PTCL/NOS, corresponding to about 60%–70% of PTCLs, cannot be further classified on the basis of morphology, phenotype, or conventional molecular studies. Clinically, PTCLs/NOS are highly aggressive lymphomas, with a poor response to therapy, and dismal overall survival (20-30%). Their pathobiology is poorly known, though recent gene expression profiling (GEP) studies have provided some hints for better understanding their pathogenesis. In particular, GEP and immunohistochemical studies on tissue-microarrays (TMAs) demonstrated PDGFRA to be systematically activated in almost all PTCLs/NOS, by nominating it as potential therapeutic target. Aims In this study, we aimed to identify the determinants of PDGFRA activation in PTCL/NOS. Specifically, we studied PDGFRA locus in order to identify possible mutations, translocations, or copy number variations and we explored the possible existence of an autocrine/paracrine loop sustaining an otherwise integer kinase. Methods The PDGFRA locus (4q1.1-4q1.3) was studied by FISH and wide-genome SNPs analysis (Affymetrix 500K Array). Direct sequencing of all PDGFRA exons and introns as well as of the promoter region was also performed in 90 cases. IHC and ELISA were adopted in order to study the expression of PDGF-A, PDGF-B and PDGF-C on tissue sections and in supernatants from PTCL/NOS cell cultures, respectively. Finally, the expression of PDGFRA and its activated (phosphorilated) form, p-PDGFRA, was assessed by IHC on TMAs, and by flow-citometry in PTCL/NOS cultured cells as well as in a FIP1L1-PDGFRApos chronic eosinophilic leukemia cell line (EOL-1) before and after the exposure to an anti-PDGF ligand neutralizing antibody (R&D System), given at various concentrations (20-40-60-80 ug/mL). Vitality assessments, proliferation/cell cycle assay (by In Situ Cell Proliferation kit, FLUOS – Roche) and evaluation of PDGFRA and p-PDGFRA were performed at 24, 48, 96 hours. A human PDGF peptide (R&D Sytems) was added to cultured cells for 6 hours to evaluate whether PDGFRA de-phosphorilation was really due to PDGF ligand remotion. Results First, FISH, SNPs analysis and direct sequencing showed preserved integrity of PDGRA locus. Thus we tested the hypothesis of an autocrine/paracrine stimulation. PDGF-A, PDGF-B and PDGF-C were found to be expressed by neoplastic cells at IHC in 93-95% of cases. In addition, PDGF-AA was found to be secreted by cultured neoplastic cells by ELISA. Notheworthy, PTCL cells secreted much more ligand than any other cell taken as control. We then tested whether PDGFRA phosphorylation was actually due to the presence of a PDGF ligand. Indeed, PTCL cells treated with anti-PDGF ligand neutralizing antibody at various concentrations showed PDGFRA dephosphorilation ranging from 30% up to 90% in a time dependent manner. Notably, the effect was specific as in EOL-1 PDGFRA phosphorylation was not modified at all. In addition, PTCL cells treated with a minimum of 20ug/mL of anti-PDGF ligand neutralizing antibody for 48h showed a 70% blockade of proliferation in comparison to untreated cells (BrdU assay). A further addition of 20 ug/ml of inhibitory antibody at 48 hours, increased the proliferation arrest up to 80% at 96 hours. Finally, the addition of a natural human PDGF peptide to cells previously treated with the anti-PDGF antibody, could restore PDGFRA phosphorylation confirming that PDGFRA de-phosphorilation was due to ligand remotion. Conclusions Taken together, our data demonstrate that PDGFRA activity is sustained by an autocrine loop in PTCL/NOS. In fact, though, in vivo, a possible additive paracrine effects mediated by reactive components cannot be excluded, we provide evidence that the phenomenon is largely due to neoplastic cells. Importantly, as PDGFRA signaling abrogation was associated to proliferation arrest, PDGFRA was confirmed as potential therapeutic target. Acknowledgments: this work was supported by Centro Interdipartimentale per la Ricerca sul Cancro “G. Prodi”, BolognAIL, AIRC, FIRB, RFO, Fondazione Cassa di Risparmio in Bologna, Fondazione della Banca del Monte e Ravenna, Progetto Strategico di Ateneo 2006, and Vanini-Cavagnino grant. Disclosures: No relevant conflicts of interest to declare.

Description: Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is a rare disease of controversial origin recently recognized as a neoplasm deriving from plasmacytoid dendritic cells (pDCs). Nevertheless, it remains an orphan tumor with obscure biology and dismal prognosis. In this study, we aimed to: 1) molecularly define the cellular counterpart of BPDCN and its relationship with other leukemias; 2) identify genes and cellular programs deregulated in the tumor; and 3) delineate novel potential therapeutic targets. To address these issues we collected and studied by gene expression profile (GEP) 27 BPDCN cases as well as 8 samples of non neoplastic pDCs. Further, a panel of samples including, myeloid precursors (MPs, N=4), lymphoid precursors (LPs, N=9), acute myeloid leukemias (AMLs, N=132), and acute lymphoblastic leukemias (ALLs, N=155) was analyzed. Validation was performed by immunohistochemistry (IHC) on tissue-microarrays, while functional experiments were carried out by using the CAL-1 cell line (derived from a BPDCN case). First, we recognized the cellular derivation of BPDCN, which proved to originate from the myeloid lineage and in particular from resting pDCs. Second, by comparing the GEP of BPDCN and resting pDCs, we identified genes and cellular programs deregulated in the tumor. Following, based on an integrated bio-informatic approach, including four different tools, we uncovered the aberrant activation of the NF-kB pathway that was confirmed in independent assays. Interestingly, among other molecules, we identified BCL2 and IRF4, two well known NFkB targets, as aberrantly upregulated in neoplastic samples and confirmed this observation by IHC. We then tested whether NFkB inhibition could represent a potential therapeutic strategy in this setting. We treated BPDCN cells ex vivowith either the proteasome inhibitor bortezomib or the selective IKKB inhibitor BMS-345541 and found them to be effective in inducing cell cycle arrest and apoptosis at relatively low dosage. By contrast, BPDCN cells turned out to be virtually insensitive to cytarabine, one of the most used drug in this condition. GEP and immunocytochemistry were then successfully used to prove that cell death was accompanied by NFkB shut-off. In conclusion, we identified a molecular signature representative of the transcriptional abnormalities of BPDCN and developed a cellular model proposing the first molecular targeted therapeutic approach in the setting of this currently incurable disease. Funding This work was supported by AIRC (IG10519 and 5xMille10007, Prof. Pileri), Centro Interdipartimentale per la Ricerca sul Cancro “G. Prodi”, BolognAIL, RFO (Prof. Pileri, Prof. Piccaluga), FIRB Futura 2011 RBFR12D1CB (Prof. Piccaluga), Fondazione Cassa di Risparmio in Bologna, Fondazione della Banca del Monte e Ravenna, Progetto Strategico di Ateneo 2006 (Prof. Pileri and Dr. Piccaluga) and by MIUR (PRIN 2011, Prof. Facchetti and Prof. Pileri). The authors have no conflicting financial interests to declare. Acknowledgments The Authors obtained the CAL-1 cell line from Takahiro Maeda (tmaeda@net.nagasaki-u.ac.jp), Department of Laboratory Medicine, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 1890 Poster Board I-913 Background. Primary myelofibrosis (PMF) is a clonal myeloproliferative neoplasm (MPN) characterised by a proliferation of predominantly megakaryocytes and granulocytes in bone marrow that in fully developed disease is replaced by fibrous tissue. At molecular level, no specific defect has been identified yet. Cytogenetic abnormalities occur in up to 30% of patients, the commonest including del(13)(q12-22), der(6)t(1;6)(q21-23;p21.3), del (20q), and partial trisomy 1q. In addition, approximately 50% of patients with PMF exhibit a single, recurrent, somatic mutation in the gene encoding the cytoplasmic tyrosine kinase Janus kinase 2 (JAK2). However, such mutation is not specific, also occurring in other MPN. Recently a couple of reports dealt with single-nucleotide polymorphism (SNP) array karyotyping of MPD, including some PMF. Importantly, such studies could identify previously uncovered genetic lesions, highlighting the importance of novel high resolution technologies for the detection of formerly unknown, cryptic aberrations. In this study we performed high resolution karyotyping by SNP oligonucleotide microarray by using the most updated Affymetrix array (Genome-Wide Human SNP Array 6.0) in 20 cases of myelofibrosis (MF) in order to identify novel cryptic genomic aberrations. Methods. DNA (500 ng) was extracted from peripheral blood cells (PBMNC) of 14 primary and 6 secondary MF patients. PBMNC were depleted from lymphocytes by magnetic beads. Briefly, CD3+ cells were labeled with anti-CD3 MoAb directly coupled to magnetic microbeads (Miltenyi Biotech), washed and subsequently purified using Mini-MACS technology. After selection, cell present in the positive (CD3) and negative (PBMNC) fractions were counted and submitted to flow cytometry analysis. DNA was processed and hybridized to the Affymetrix SNP arrays 6.0 as for manufacturer instruction. A whole-genome copy number variation (CNV), genotyping, loss of heterozygosity (LOH) and uniparental disomy (UPD) analyses were performed using the Partek Suite 6.0. Ten lab-specific as well as 90 HapMap samples relative to Caucasian healthy donor were used as control reference. Genomic abnormalities were defined as recurrent when occurring in at least 25% of cases. JAK2 mutational status was assessed as reported, by alle-specific PCR. Clinical information and complete follow up were retrieved for all cases. Direct sequencing, FISH, qPCR and immunohistochemistry (IHC) has been chosen for validation. Results. In all patients we could detect several CNV. The median number of CNV was 60 (range, 34-72), including 46 amplifications (A) and 14 deletions (D). All commonest previously described abnormalities were detected. In addition, several formerly uncovered recurrent lesions were identified, mainly involving 1p, 1q, 2p, 4p, 4q, 5q, 6p, 6q, 7q, 8p, 9q 10q, 11p 11q, 12p, 14q, 15q, 16p, 16q, 17q, 18q, 19q, 20p, 22q. The median size of such CNV was 424,582 Kbp (1,379 Kbp-71,277 Mbp). We then compared JAK2+ vs. JAK2− cases. Of note, we found numerous definite aberrations (A or D) distinguishing the two groups and specifically affecting 16q23.1, 1p36.13, 3q26, 14q13.2, 5q33.2, 6q14.1, 7q33, 8p23.1, and 9p11.2. Grippingly, several genes of potential interest for PMF pathogenesis were identified within the involved loci, including RET, SCAPER, WWOX and SIRPB1. Among others, the product of such genes has been selected for validation by IHC. Similarly, many miRNA were recognized, which may deserve further investigation. Conclusions. By using a newly developed highly sensitive array we identified novel cryptic lesions in patients affected by MF. Future studies on larger series, as well as functional analyses will definitely assess their role in the pathogenesis of the disease. Of note, consistent differences were recorded in JAK2+ vs. JAK2−, supporting the hypothesis of different genetic mechanisms occurring in the two sub-groups. Acknowledgments: this work was supported by AIL Pesaro Onlus, Centro Interdipartimentale per la Ricerca sul Cancro “G. Prodi”, BolognAIL, AIRC, FIRB, RFO, Fondazione Cassa di Risparmio in Bologna, Fondazione della Banca del Monte e Ravenna, Progetto Strategico di Ateneo 2006.*GV and MRS equally contributed to this work. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 3662 The differential diagnosis among the commonest peripheral T-cell lymphomas (PTCLs) (i.e. PTCL not otherwise specified, NOS; angioimmunoblastic T-cell lymphoma, AITL; and anaplastic large cell lymphoma, ALCL) is difficult, the morphologic and phenotypic features being largely overlapping. Noteworthy, recent international studies indicated significant differences in their clinical behavior as well as concerning the presence of potential therapeutic targets. We performed whole genome gene expression profiling (GEP) of PTCLs aiming to identify molecular signatures able to improve their diagnosis. We studied 95 PTCLs, including 73 PTCLs/NOS, 12 ALCLs (6 ALK+ and 6 ALK-), and 10 AITLs. All tissue samples were formalin-fixed and paraffin embedded (FFPE). GEP was performed by Illumina Whole Genome DASL Assay. First, we documented the efficiency of GEP from FFPE tissues by comparing the mRNA levels and the presence of the corresponding protein, including expressed (i.e. CD3) and not expressed (i.e. BCL10) molecules. Secondly, we tried to discriminate different PTCLs basing on their GEPs. By dividing a training (N=47) and a test set (N=48), we found 2 signatures able to differentiate PTCL/NOS vs. AITL and PTCL/NOS vs. ALCL ALK-. Specifically, in the test set the sensitivity (ST) and specificity (SP) of the assays were 100% – 80% (PTCL/NOS vs. AITL) and 100% – 100% (PTCL/NOS vs. ALK- ALCL) (Table 1). Accordingly, the positive (PPV) and negative (NPV) predicting values for the identification of PTCL/NOS were 0.92 and 1 (vs. AITL) and 1 and 1 (vs. ALK- ALCL) (Table 1).Table 1.Accuracy of GEP based signature in differentiating PTCL subtypesSTSPPPVNPVTraining setPTCL/NOS vs. AITL100%80%0.921PTCL/NOS vs. ALK-ALCL100%100%11Test setPTCL/NOS vs. AITL92.50%100%10.77PTCL/NOS vs. ALK-ALCL92.50%100%10.67Validation setPTCL/NOS vs. AITL85%86%0.920.76PTCL/NOS vs. ALK-ALCL96%73%0.960.73 Interestingly, the identified genes represented relevant functional pathways differentially regulated in the 3 tumour types, including protein kinase cascade, proliferation, and cell cycle. When applied to the test set of cases, the assay correctly classified 37/40 PTCLs/NOS (92.5%), 5/5 AITLs, and 3/3 ALK- ALCLs. Finally, we tested our signatures on 133 independent PTCL cases (including 78 PTCL/NOS, 43 AITL, and 12 ALK- ALCL) for which GEP data were available on the GEO database and were originally obtained from fresh/frozen tissues. Interestingly, we could efficiently recognize PTCL/NOS cases vs. AITLs (ST, 85%; SP 86%; PPV 0.92; NPV 0.76) and vs. ALK- ALCLs (ST 96%; SP 73%; PPV 0.96; NPV 0.73). In conclusion, we successfully generated for the first time GEP from routinary FFPE PTCL samples, identifying molecular signatures potentially useful for the clinical practice and, specifically, for the differential diagnosis of PTCL types. Disclosures: No relevant conflicts of interest to declare.

Description: Burkitt lymphoma (BL) is classified into 3 clinical subsets: endemic, sporadic, and immunodeficiency-associated BL. So far, possible differences in their gene expression profiles (GEPs) have not been investigated. We studied GEPs of BL subtypes, other B-cell lymphomas, and B lymphocytes; first, we found that BL is a unique molecular entity, distinct from other B-cell malignancies. Indeed, by unsupervised analysis all BLs clearly clustered apart of other lymphomas. Second, we found that BL subtypes presented slight differences in GEPs. Particularly, they differed for genes involved in cell cycle control, B-cell receptor signaling, and tumor necrosis factor/nuclear factor κB pathways. Notably, by reverse engineering, we found that endemic and sporadic BLs diverged for genes dependent on RBL2 activity. Furthermore, we found that all BLs were intimately related to germinal center cells, differing from them for molecules involved in cell proliferation, immune response, and signal transduction. Finally, to validate GEP, we applied immunohistochemistry to a large panel of cases and showed that RBL2 can cooperate with MYC in inducing a neoplastic phenotype in vitro and in vivo. In conclusion, our study provided substantial insights on the pathobiology of BLs, by offering novel evidences that may be relevant for its classification and possibly future treatment.