ALBERT

Description: INTRODUCTION In newly diagnosed Multiple Myeloma (MM) patients (pts), Copy Number (CN) losses of chromosome 17p13, carrying the TP53 tumor-suppressor gene, are strong predictors of poor outcomes. On the contrary, the prognostic relevance of TP53 mutations at the onset of the disease is less clear, due to the very limited frequency of clonal lesions, as revealed by Sanger sequencing. To address this poorly investigated issue, we used an ultra-deep sequencing (UDS) approach to characterize the TP53 structural architecture in both newly diagnosed and relapsed MM pts and to assess the prognostic role and evolution over time of small TP53 mutated sub-clones. SAMPLES AND METHODS A cohort of 99 newly diagnosed MM pts treated up-front with bortezomib-based regimens and autologous stem cell transplantation, was included in this molecular study. In 29 cases, samples were collected both at diagnosis and at relapse(s). DNA was obtained from CD138+ highly purified plasma cells. TP53 gene mutational status was analysed by using an amplicon-targeted UDS approach (GSJ, 454 Roche Life Sciences). In order to discriminate between low frequency sub-clonal TP53 variants and sequencing errors, sequencing raw data were filtered according to cut-off values based on different ranges of sequences' coverage depth. Additional filters were also applied, based on both quality and biological cut-offs, to obtain a final confident list of variants. Analysis of CN alterations (CNAs) was performed by SNPs array and results analysed with ChAS software. RESULTS With a median coverage of 1386X, a list of 129 correctly called TP53 variants (either missense, or nonsense or splice ones), including 20 INDELs, was detected. Only deleterious and N/A variants (according to SIFT classification) were included in the list. Most newly diagnosed MM pts (55%) carried at least one TP53 sub-clonal variant (on average 1.08 variants per pts), with 45/99 (45%) carrying non-mutated TP53. Pts carrying TP53 sub-clonal variants bared also TP53 CN hemizygous losses (20%), CKS1B gains (56%) and cdkn2c losses (14%). According to TP53 sub-clonal mutational load, pts were stratified in two sub-groups, including 28 pts with ≥2 (high load) and 71 with

Description: Background Although remarkable advances have been achieved in MM therapy, mainly thanks to the introduction of novel-agent-based regimens, the disease remains incurable. Neoplastic CD138+ plasma cells are the hallmark of MM: both their expansion in the bone marrow (BM) and the production of monoclonal immunoglobulin (Ig) are responsible for the clinical manifestation of the disease. However, the existence of a Myeloma Propagating Cells (MPCs) has been proposed as a major cause of MM drug-resistance, leading to relapse. Several studies support the hypothesis that MPCs are phenotypically close to memory B cells residing in the CD138- compartment; however, very little is known concerning their molecular characteristics. Here we present an extensive molecular characterization of clonotypic CD19+ B cells clones obtained from newly diagnosed MM patients (pts), in order to recognize biological pathways possibly explaining the malignant clone’s persistence. Methods CD138+ and CD138- cell fractions were collected from BM and peripheral blood (PBL) of 50 newly diagnosed MM pts. CD19+ B cell and CD27+ memory B cell populations were isolated from CD138- cell fraction. Clonogenic assays were performed by plating cell fractions obtained from RPMI-8226 and NCI-H929 cell lines. The molecular characterization included: IgH gene rearrangement Sanger sequencing; analysis of the whole spectrum of genomic aberrations and gene expression profiling, by Affymetrix 6.0 SNPs array and HG-U133 Plus 2.0 microarray, respectively. Results Clonogenic assays showed that CD138- cells, plated on conditioned media, were able to form colonies after two weeks of culture more efficiently than CD138+ cells. By VDJ gene rearrangement sequencing, a clonal relationship between the CD138+ clone and the memory B ones was confirmed. SNPs arrays showed that both BM and PBL CD138+ cell fractions carried exactly the same genomic macro-alterations. On the contrary, in the CD138-19+27+ cell fractions from BM and PBL any macro-alteration was detected, whereas several micro-alterations (median number per sample: 32 amplifications and 16 losses, range: 8-122 Kb, average markers per region: 50) unique of the memory B cells clone were highlighted. An enrichment analysis revealed the involvement of genes affected by losses (17 genes) in both DNA repair mechanisms and transcriptional regulation and the involvement of genes affected by gains (46 genes) in both the negative regulation of apoptosis and the angiogenesis. Interestingly, KRAS, WWOX and XIAP genes, renown to be involved in MM pathogenesis, are located in the amplified regions in the immature cells. Moreover, several LOH regions were described, which covered at least 106 tumor suppressor genes involved in MM and leukemia (including TP53, CDKN2C and RASSF1A). Transcriptome profiles analysis of the CD19+ cell fractions highlighted pathways suggesting a possible involvement of immature cells in MM pathogenesis. The gene expression profiles of 20 MM CD19+ cells samples (12 from PBL, 8 from BM) were compared both to their normal counterpart and to the mature CD138+ cell fractions. In particular, unsupervised analysis by hierarchical clustering discriminated the differential expression of 11480 and 11360 probes in the PBL and BM CD19+ clones, respectively (2; FDR=0,05; p

Description: Background Multiple Myeloma (MM) is a biologically complex disease, whose genetic plasticity favors the coexistence of genetically heterogeneous subclones, selected in a Darwinian fashion throughout the disease course. Based on a disease evolution model, it is likely that therapy acts like an evolutionary bottleneck, applying a selective pressure on the clones' genomic background. However, a formal correlation between the modulation of intra-clonal heterogeneity and the treatment has yet to be demonstrated. Aim To explore the existence of different clonal evolution patterns in MM, eventually driven by therapeutic selective pressure, we longitudinally analyzed a cohort of patients (pts) by using a high-throughput technology, able to finely dissect the genomic intra-clonal changes occurring in each pts, across disease progression. Patients and methods Thirty-three pts with symptomatic MM were included in this study. Most pts (28) were up-front treated with regimens including a proteasome inhibitor (PI), either in combination with an immunomodulator (IMiD) (thalidomide or lenalidomide) or with cyclophosphamide. The remaining pts (5) were up-front treated with IMiDs-based regimens, not including PI. For each pts, paired bone marrow samples were collected both at diagnosis and at relapse. SNPs array analyses were performed on the CD138+ enriched cell fractions (Affymetrix 6.0 and CytoscanHD) and data were analyzed with ChAS v3.1 and Nexus software, to obtain Copy Number Alterations (CNAs) results. Results: The genomic landscape's modulations were evaluated in details in each pair of samples included in the study: by one side, by monitoring the variations of macro CNAs types; by the other, by focusing on changes of CNAs frequencies, as observed in 27 genes of interest (selected according to their recognized role in MM pathogenesis). Both approaches were consistent in highlighting three major evolution patterns: in 7/33 (21%) pts, the genomic background at relapse was almost identical to that of diagnosis, suggesting that no evolution affected the diagnosis' genomic architecture. In 13/33 (39%) pts, an overall increase in the frequencies of the same CNAs as observed at diagnosis was detected at relapse, suggesting a linear evolution of the diagnosis' predominant clone. Finally, in 13/33 (39%) pts, either increased or decreased frequencies of several CNAs, as well as several differences in the CNAs type's prevalence were observed in samples collected at relapse, as compared to those collected at diagnosis, suggesting a branching evolution of sub-clones not-detectable at diagnosis. Of interest, even if an overall increase in the median number of CNAs was observed in the CD138+ cell fractions collected from the whole population of newly diagnosed and relapsed pts (226 vs 507, respectively) - supported by acquisition of CNAs either commonly described as secondary genomic events (i.e. del17p13, amp1q21, del1p23), or associated to the resistance to bortezomib (i.e. del8p21) - nevertheless any peculiar CNAs resulted significantly prevalent in the three identified subgroups of pts. The subgroup of pts who underwent a branching evolution was characterized by a high rate (92%) of achievement of VGPR or better quality of response to upfront therapy, regardless of incorporation of PI into induction regimens. On the contrary, although pts who experienced either linear or no evolution were mostly treated with highly effective PI-based triplet combinations (19/20 pts), the rate of VGPR in this subgroup was only 20% and PR or SD were observed in 9 and 7 pts, respectively. Finally, the median time to first progression of this subgroup of pts was significantly shorter as compared to that of pts with branching evolution (24 vs 35 months, range 4-41 and 7-123 months, respectively, p=0,01). Conclusion The genomic architecture of a subgroup of relapsed MM pts, who were up-front responsive to new drugs-based combination therapies, resulted overall different from that of diagnosis, suggesting a branching evolution of the disease, sustained by the shrinking of the most prevalent clone (therapy-sensitive), as well as by the expansion of subclones (therapy-resistant) not already evident at diagnosis. This observation raises the question whether re-treatment of relapsed pts should be appropriate in the case of branching evolution. Acknowledgments: AIRC (MC), Fondazione Berlucchi (CT), FUV (EB). Disclosures Cavo: Takeda: Honoraria; Amgen: Honoraria; Bristol-Myers Squibb: Honoraria; Celgene: Honoraria, Research Funding; Janssen: Honoraria, Research Funding.

Description: Introduction. The role of TP53 tumor-suppressor gene in mediating cellular basic tumor suppressive mechanisms is crucial: its utmost importance is underscored by a 50% mutational rate among most human cancers. In MM, TP53 is rarely reported as being affected either by deletions or mutations, even if chromosome (chr) 17p13 copy number (CN) loss defines a samples subgroup with a particularly poor prognosis. Here we aim at retrospectively analyzing by Next Generation Sequencing (NGS) the TP53gene inactivation in newly diagnosed MM, assessing the mutational events’ frequency and their clinical impact. Samples and methods. A cohort of 92 MM, receiving up-front velcade (vel)-based regimens, followed by autologous stem cell transplantation, was included in this molecular study. Response to therapy was evaluated according to the IMWG criteria. DNA was obtained from CD138+ highly purified plasma cells and profiled by SNPs array (Affymetrix). TP53gene mutational status was analysed employing an amplicon-targeted deep NGS approach (GSJ, 454 Life Sciences). A novel upcoming NGS data analysis software was employed to detect variants from raw sequencing data and to compare them to IARC database. As a validation, 30 randomly chosen samples have been re-sequenced. p53 activity was evaluated by p-p53 and p53 immunoblot assays. Results. Ultra-deep TP53coding sequence analysis (median depth: 1060) highlighted the presence of a median of 1.8 variants per samples in 73/92 (79%) MM: a total of 131 nucleotide substitutions emerged, with variants allele frequencies (VAF) ranging from 1 to 99% (median 1.3%). The polymorphism c.215C〉G (validated as SNP1042522) was the most prevalent detected variant, recurring in ~40% of the analysed samples. TP53variants were assessed according to the SIFT classification, in order to identify either neutral or deleterious mutations, on the basis of their predicted effect at amino acid level. 36 cases carried 42 deleterious variants (VAF=1.2%): they were mostly missense substitutions affecting mainly the DNA binding domain; most deleterious variants (67%) were predicted as non-functional at protein level. 9 samples carried 21 neutral variants (VAF=1.1%), which were mostly missense mutations, again mainly affecting the DNA binding domain. Finally, 28 samples carried 68 SIFT unclassified variants – among which several polymorphisms were counted – (VAF=94.6%). Of these, 29 were silent substitutions affecting the DNA-binding domain, 32 were missense variants affecting the SH3-like/Proline- rich domain. To assess the genomic background of deep-sequenced MM, SNPs arrays were performed on 83/92 samples: 14/83 (17%) carried a TP53 CN loss (1N) on chr17p13.1. The incidence of TP53 hemizygous deletion was higher among cases carrying deleterious mutations, as compared to cases carrying either neutral or unclassified mutations (25%, 12,5% and 12%, respectively). 2/17 non-mutated samples carried TP53 hemizygous deletion, as well. Interestingly, an Rb1 tumour suppressor gene CN loss on chr13q14.2 significantly characterized samples carrying either mutated (deleterious variant) or deleted TP53 (p=0.006). The activation state of p53 was evaluated by p-p53 immunoblotting assay, showing the absence of phosphorilation either in 3 TP53 deleted cases, or in 3 cases carrying deleterious variants; on the contrary p-p53 was observed in 4 non-mutated cases. Finally, clinical correlations were performed on 81 MM, for whom clinical data were available, showing that the presence of either TP53 hemizygous deletions or at least one TP53 deleterious variant was more likely associated to the response to vel-based induction therapy (frequencies of ≤partial response were 35% and 64% among samples carrying or not impaired TP53, respectively; p=0.05). Conversely, the frequency of progression events was slightly higher among MM carrying impaired TP53 (69% vs. 50%). Conclusions. The analysis by ultra-deep next generation sequencing of TP53 coding sequence in a cohort of newly diagnosed MM highlighted an unexpected, still un-explored, high rate of TP53 variants, recurring with a wide range of frequencies among samples. The impact of TP53 damage on MM disease course has to be confirmed in randomized clinical trial. Acknowledgments: Roche Diagnostics for supporting us in the realization of this project; ELN, AIL, AIRC, PRIN, progetto Regione-Università 2010-12 (L. Bolondi), FP7 NGS-PTL project. Disclosures Haferlach: MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Cavo:Celgene: Consultancy, Speakers Bureau; Janssen: Consultancy, Speakers Bureau; BMS: Consultancy, Honoraria; Millenium: Consultancy, Honoraria; Onyx: Honoraria. Martinelli:Novartis: Consultancy, Speakers Bureau; BMS: Consultancy, Speakers Bureau; Pfizer: Consultancy; ARIAD: Consultancy; Roche: Honoraria.

Description: INTRODUCTION. Although remarkable advances have been reported in Multiple Myeloma (MM) therapy, mainly due to the introduction of novel agents, the disease remains incurable in most of the patients. The iperactivation of self-renewal mechanisms, like Hedgehog (Hh) pathway, which controls the refuel of the tumor clone, might be critical to disease recurrence. Whilst several studies suggestthatHh pathway is activated in the putative CD138- Myeloma Propagating Cells (MPCs), it is likely that also terminally differentiated CD138+ plasma cells might contribute to drug resistance, by reverting to an immature phenotype. AIM. In order to dissect the role played by Hh pathway in different MM cells compartments, and to evaluate the impact of Hh pathway expression on patientsÕ clinical outcomes, a high-throughput molecular characterization was employed to explore the transcriptomic and genomic profiles in both CD138+ plasma cells and CD138-19+ B cells progenitors obtained from newly diagnosed MM patients. PATIENTS AND METHODS. The study included a cohort of 126 patients, homogenously treated with bortezomib-based regimens and ASCT, who were randomly included in a training set and a test set. For each patient, the CD138+ plasma cell fraction was isolated by immunomagnetic beads method; CD19+ B cells were isolated in 18 patients. Gene expression profiling (GEP) (HG U133 Plus 2.0 chip) and genomic analysis (SNP 6.0 chip) were performed on Affymetrix platform. dChip analysis software was used to perform GEP clustering. Expression data were analyzed by Ingenuity Pathway Analysis software and were validated by Western Blot assays. Copy number analysis was carried out using Nexus Copy Number software. RESULTS. The expression of Hh pathway genes resulted deregulated in both CD138+ and CD19+ cells, as compared to their normal counterparts. By unsupervised hierarchical clustering, an Hh signature of 10 genes - SHH, IHH, DHH, SMO, PTCH1, PTCH2, SUFU, GLI1, GLI2 and GLI3 - was identified, and was able to significantly cluster patients in two subgroups: cluster 1 included 39 patients while 37 were included in cluster 2. Clustering robustness was validated in an independent cohort of 50 patients (test set), of whom 31 were assigned to cluster 1 and 19 to cluster 2. An overall significant activation of Hh pathway was shown in cluster 2, as compared to cluster 1. Of note, the Hh pathway was down regulated in CD19+ B cells obtained from patients included in cluster 2, while it was overexpressed in cluster 1 patients. Western blots on both cell fractions confirmed this opposite Hh genes behavior. Peculiar genomic and transcriptomic profiles characterized patients included in clusters 1 and 2: indeed, a higher genomic instability (e.g. higher frequencies of both t(4;14) and del(17p)) was demonstrated in CD138+ plasma cells from cluster 2 patients and, at least 5 known tumor suppressor genes, such as RB1, BRCA2, PDX1, FOXO1 and TP53 were included in deleted regions. Conversely, cluster 1 patients were mainly characterized by hyperdiploid karyotypes. The more aggressive phenotype of cluster 2 patients was confirmed by an overall deregulation of cell adhesion processes (CD44, LIMS1, COL4A2, CTGF, COL1A1, FN1), increased proliferation (MYCBP, IL22, SDPR, SOX2, SOX6) and impaired DNA repair mechanisms (SP1, SMARCD3, FOXA3). Hh pathway activation significantly influenced patientsÕ outcome, since those included in cluster 2 had a shorter PFS and OS compared to cluster 1. In fact, the 5-year PFS estimates were 31% vs 56% (p=0.0062), whereas the OS probabilities were 66% and 83%, respectively (p=0.0071) (Fig.1,2). Of note, both hazard ratios for PFS and OS were doubled in patients included in cluster 2, as compared to patients included in cluster 1. Finally, multivariate analyses confirmed that being included in cluster 2 was an independent prognostic factor for both PFS and OS, along with del(17p) and ISS 3 (Tab. 1). CONCLUSION. Sorts of Òying -yang Ó effect of Hh pathway between mature CD138+ plasma cells and immature CD138-CD19+ MPCs could be hypothesized, where two alternate Hh-driven subtypes of MM at diagnosis correlated well with patientsÕ outcomes. Stratification of patients according to their molecular background might help the fine-tuning of future clinical studies. Supported by Regione-Universita 2010-12 (L. Bolondi), FP7 NGS-PTL project. Disclosures Zamagni: Celgene Corporation: Honoraria, Speakers Bureau; Janssen Pharmaceuticals: Honoraria, Speakers Bureau; Amgen: Honoraria, Speakers Bureau. Martinelli:AMGEN: Consultancy; BMS: Consultancy, Speakers Bureau; Ariad: Consultancy; Novartis: Consultancy, Speakers Bureau; MSD: Consultancy; Pfizer: Consultancy; ROCHE: Consultancy. Cavo:Janssen-Cilag, Celgene, Amgen, BMS: Honoraria.

Description: Introduction: Novel array-based technique-single-nucleotide polymorphism (SNP) microarray can detect cytogenetic lesions mostly involving structural alterations with losses or gains of chromosomic material. These abnormalities are predictive of response and can help define therapeutic strategies. SNP microarray can also detect copy-neutral loss of heterozygosity (CN-LOH), which has a described role in Acute Myeloid Leukemia (AML) by inducing oncogene duplication, tumor suppressor inhibition and epigenetic reprogramming. Aim: To improve conventional cytogenetic analysis and identify new genes relevant to leukemogenesis by SNP array-based genotyping. Materials and Methods: We analyzed 279 AML patients (pts) at diagnosis by SNP Array 6.0 or Cytoscan HD Array (Affymetrix). Thirty-four samples were also analyzed by Whole Exome Sequencing WES (HiSeq 2000,Illumina). SNP Array data were analyzed by Nexus Copy Number™ v7.5 (BioDiscovery) and R Development Core Team, while WES data were analyzed by GATK and MuTect. Results: Copy Number Alterations (CNAs) were scattered across all chromosomes (chrs). All pts showed CNA events: 44.4% of CN gain, 21% of CN loss and 34.6% of CN-LOH. Single copy gains mainly affected chrs X, 1, 2, 4, 9 and 8. Duplications occurred at chrs 2, 3, 5 and 14. Heterozygous loss events were detected in chrs 3, 5 and 14, while regions of deletion were located in chrs 6, 7 and 22. The CN-LOH event was the most common event and involved chrs 1, 2, 3, 4, 5, 6 and 8. We studied the deletome profile in our cohort (Fig. 1) pf pts in order to define the minimal common deletion region. SNP array analysis showed that several genes were preferentially deleted, including ADAM5 (12,9%), PHF6 (12,2%), AGPS (10%), SOX6 (7,9%), WT1 (6,5%), CRLF2 (5,4%) and LRRK1 (4,3%); while the genes preferentially amplified were GPC3 (70,25%), FLT3 (44,8%), FGF13 (36,2%), KIT (31,54%), AFF2 (31,5%), ETS1 (26,52%), MITF (22,2%), CASK (16,5%), CDY1 (14,33%), MECP2 (14,33%), FOXP2 (14%) and SMAD4 (12,9%). Single-copy losses and deletions were enriched (p