ALBERT

Description: The role of the HOX gene family in leukemia development has been extensively studied. However, these studies have focused almost exclusively on the potential oncogenic role of HOX gene family members. In contrast to the oncogenic function often attributed to HOX genes, our studies have identified several HOX gene family members as candidate tumor suppressor genes and shown that inactivation of HOX genes, particularly HOXA4, is associated with poor prognosis. We have used multiple quantitative methylation assays to search for epigenetic inactivation of HOX genes in adult and childhood leukemia. In both adult myeloid and lymphoid leukemia two members of the HOXA cluster (HOXA4 and A5) were found to be frequently inactivated by promoter hypermethylation (26–64% of cases were hypermethylated). In contrast, a further 12 HOXA, B and C cluster genes were found to be essentially devoid of hypermethylation (except HOXA6 in CLL, where 34% of samples exhibited hypermethylation). HOXA4 and HOXA5 were also frequently inactivated in childhood ALL and AML (39–79% of samples). However, in contrast to the adult leukemias, all but one of the additional HOX genes analyzed were also found to be targets for hypermethylation in both ALL and AML (4–26% of samples), suggesting that HOX genes are differentially regulated in childhood versus adult leukemia. Hypermethylation of HOX genes (HOXA4, HOXA5 and HOXA6) was associated with loss of expression of the corresponding gene. Expression analysis also suggests that interaction between different HOX genes may be crucial. In normal karyotype AML samples, those expressing of high levels of HOXA9, but not those with low HOXA9 expression, were associated with invariable HOXA4 hypermethylation (p=0.01). Interestingly HOXA4 hypermethylation also correlates with poor prognosis in all types of leukemia tested. Hypermethylation of HOXA4 correlates with progression to blast crisis (p=0.007) and poor response to imatinib in CML (p=0.04), with cytogenetic status in AML (33%, 72% and 100% in good, intermediate and poor prognostic groups respectively, p=0.0004) and with IgVh mutational status (p=0.003) and poor survival in CLL (median survival 159 versus 199 months in hypermethylated and non hypermethylated patients, respectively). Furthermore transfection of a HOXA4 expressing construct into a CML blast crisis cell line results in re-expression of markers of myeloid differentiation, suggesting that loss of HOXA4 is functionally relevant in leukemic cells. These results indicate that aberrant epigenetic regulation of HOXA4, and indeed other frequently inactivated HOX genes such as HOXA5 and HOXA6, may play a key role in the development of multiple types of leukemia. Thus co-ordinated up and down regulation of expression of HOX gene family members may be crucial in the leukemogenic process.

Description: Abstract 467FN2 The Ataxia Telangiectasia Mutated (ATM) gene located at 11q23 plays a central role in DNA damage response. Deletion of 11q23 occurs in approximately 20–30% chronic lymphocytic leukemia (CLL) cases and in 30% of these exhibit mutations in their remaining ATM allele. A smaller proportion of CLL tumours exhibit the presence of ATM mutations in the absence of 11q deletion. We have previously shown that ATM mutations are associated with a shorter overall survival from diagnosis and treatment free survival (TFS) in an unselected CLL cohort. Furthermore, several prospective and retrospective studies, including the UKCLL4 trial, show an inferior overall survival and progression free survival after therapy (OS and PFS) in cases with an 11q deletion treated with alkylating agents and/or purine analogues. The entire ATM coding region was screened for sequence changes using high-performance liquid chromatography and sequencing and known polymorphisms excluded. Screening was performed on 238/777 cases enrolled in the UKCLL4 trial, including all available cases with 11q deletions, with the remaining cases being randomly selected. Other than significant enrichment for patients with 11q deletion (p

Description: Whilst the spectrum and clinical significance of gene mutations and immunogenetic features in common mature B-cell malignancies is well established, their incidence, biological and clinical importance in splenic marginal zone lymphoma (SMZL) remains uncertain. Accordingly we screened 175 well-characterised SMZL cases for mutations in 768 genes (Haloplex Target Enrichment System) with a known or postulated role in B-cell physiology, B-cell malignancies in general and SMZL pathophysiology in particular. After sequencing (HiSeq 2000) we achieved a mean depth across our gene panel of 297-fold (range 128-702x), with more than 85% of all bases covered at 〉50-fold. After comprehensive filtering, 1,374 single nucleotide variants and insertions/deletions were identified. 221 genes were recurrently mutated at a gene frequency of 2-16% [n=2-28]. Sanger sequencing confirmed 86/86 selected variants in our recurrent genes, and showed 99% concordance between our Haloplex and Sanger sequencing of NOTCH2 exon 34, which was performed in all patients. Comprehensive validation of both germ-line Haloplex [n=18 patients] and Sanger sequencing established the sensitivity and specificity of our approach, and confirmed the biological importance and somatic origin of the genes described herein. To identify biologically relevant genes, we employed MutSigCV analysis, an algorithm that identifies significantly mutated genes by accounting for background mutation rate, DNA replication time and the gene size. 18 mutated genes were identified with TP53 [n=28], KLF2 [n=21], MYD88 [n=12], NOTCH2 [n=17], TNFAIP3 [n=13] and CCND3 [n=15] being the most significant; genes that encode components of pathways important in the regulation and differentiation of mature B-cells were also identified, including CREBBP [n=9], MAP3K6 [n=5], KDM2B [n=7], SETD1B [n=6], TRAF3 [n=9], ARID1A [n=10], BIRC3 [n=3], BCL10 [n=5], BTG1 [n=3], ATM [n=10], NFKBIE [n=4] and DDI1 [n=4]. Then, we searched for significant pairwise gene correlations and mutually exclusive relationships between our mutated genes demonstrating the following: (1) independent events, such as MYD88, where a mutation is invariably observed as an isolated event; (2) cancer drivers that have a similar proportion of co-occurring and mutually-exclusive relationships, such as NOTCH2, TP53, TNFAIP3 and CREBBP, and (3) genes such as KLF2, CCND3 and ARID1A that have proportionally more co-occurring relationships, thus suggesting a synergistic function to promote tumorigenesis. Finally, we studied clonal evolution, by differentiating between early, clonal events, and later, subclonal mutations (ABSOLUTE algorithm), and we were able to classify clonal or subclonal mutation in 6/18 of our MutSigCV genes. Paradigmatically, we observed that all the CREBBPmutations were fully clonal. Amongst our most novel findings was KLF2, or Krüppel-like factor 2, mutations that were distributed across the entire protein, with a cluster in the C2H2 domain and were all somatically acquired. All mutations tested were clonal, significantly associated with del(7q) (P=0.001), IGHV1-2*04 gene usage (P

Description: Abstract 757 Chromosomal deletions involving 13q14 are the commonest genetic abnormality in chronic lymphocytic leukemia, occurring in ~60% of cases and are associated with a favourable clinical outcome when identified as a sole abnormality. A minimally deleted region (MDR), found in most cases, encompasses DLEU2, DLEU1, and miR15a/16-1. However, deletions this small do not occur in all patients and are a simplification of the impact larger heterogeneous deletions have during carcinogenesis. We have previously shown that large 13q14 deletions are associated with disease progression (Strefford et al, ASH 2009, 114(22), 671). To further characterize this abnormality and identify putative cancer genes, we report an updated analysis of our Affymetrix SNP 6.0 genomic profiling study of 224 patients. These patients were subdivided into three cohorts: Cohort I samples were taken at disease presentation, from patients with either stable disease for 〉5 years (n=38) or progressive disease within 3 years (n=25). Cohort II comprised 64 unselected patient samples taken at disease progression. Cohort III consisted of samples taken at enrollment to the UK CLL4 treatment trial from patients treated with fludarabine and cyclophosphamide, sub-divided based on complete (n=49), partial (n=40) or no response (n=8) to treatment. In total 205 copy number alterations targeted 13q14 in 132 cases, facilitating the identification of a MDR encompassing DLEU2 and the miR15a/16-1 cluster in 119 cases. The remaining cases had partial loss of the MDR, including (n=11) or excluding (n=2) miR15a/16-1. Although deletion size and location was heterogeneous (0.13-96.2Mb), a 210kb proximal breakpoint cluster region (p-BCR) was identified, targeting TRIM13, KCNRG and exons 7–11 of the DLEU2 gene (n=31), whilst a 210kb distal BCR (d-BCR) targeted RNASEH2B and GUCY1B2 (n=46). Breakpoints in these regions frequently targeted DNA repeats, specifically short (SINE) and long interspersed nuclear elements (LINE) in the p- and d-BCR, respectively. Based on size and location, we defined two deletion classes; both encompassed the MDR, whilst class I deletions were

Description: NOTCH1 and SF3B1 mutations have been previously reported to have prognostic significance in chronic lymphocytic leukemia but to date they have not been validated in a prospective, controlled clinical trial. We have assessed the impact of these mutations in a cohort of 494 patients treated within the randomized phase 3 United Kingdom Leukaemia Research Fund Chronic Lymphocytic Leukemia 4 (UK LRF CCL4) trial that compared chlorambucil and fludarabine with and without cyclophosphamide in previously untreated patients. We investigated the relationship of mutations in NOTCH1 (exon 34) and SF3B1 (exon 14-16) to treatment response, survival and a panel of established biologic variables. NOTCH1 and SF3B1 mutations were found in 10% and17% of patients, respectively. NOTCH1 mutations correlated with unmutated IGHV genes, trisomy 12, high CD38/ ZAP-70 expression and were associated with reduced overall (median 54.8 vs 74.6 months, P = .02) and progression-free (median 22.0 vs 26.4 months, P = .02) survival. SF3B1 mutations were significantly associated with high CD38 expression and with shorter overall survival (median 54.3 vs 79.0 months, P 〈 .001). Furthermore, multivariate analysis, including baseline clinical variables, treatment, and adverse prognostic factors demonstrated that although TP53 alterations remained the most informative marker of dismal survival in this cohort, NOTCH1 (HR 1.58, P = .03) and SF3B1 (HR 1.52, P = .01) mutations have added independent prognostic value.

Description: Abstract 658 In CLL, 11q23 deletion (encompassing the ATM gene) predicts a poor response to initial treatment with DNA damaging agents that can be ameliorated by the addition of anti-CD20 antibodies. 30–40% of 11q-deleted CLL cases have an inactivating ATM mutation on the remaining allele. In the UK CLL4 trial, bi-allelic ATM abnormalities were associated with a shorter PFS than those with mono-allelic ATM loss. The additional observation of a shorter PFS in cases with mono-allelic ATM loss compared to those with an ATM mutation in the absence of an 11q deletion suggests a potential pathogenic role for other genes within the 11q deleted region. A recent study provides evidence that loss and/or mutation of BIRC3, located at 11q22, that encodes a negative regulator of non-canonical NFkB signalling, may be an independent marker of poor outcome in CLL. In order to provide more data on the relative frequency and clinical significance of ATM and BIRC3 loss and mutation, we have screened a cohort of previously untreated patients (n=264) enriched for cases with an 11q23 deletion detected by FISH (n=80) using SNP6.0 profiling (n = 264) and HRM-PCR/DHPLC & sequencing of the ATM (all exons; n=130), and BIRC3 (mutation hotspots: exons 5,7,10; n=258) genes. SNP6.0 copy number analysis identified 112 deletions on chromosome 11 in 76 cases, of which 69 had deletions including ATM (27.8Mb; 521kb–58Mb). Although a 416Kb Minimally Deleted Region (MDR) was defined (107.5–107.9Mb; hg18) containing 6 genes including ATM, in only 4% of 11q-deleted cases was a deletion confined to the MDR and invariably 11q23 deletions were much larger. The entire BIRC3 gene was only deleted in cases with ATM loss, and always within a single deletion event encompassing both genes. BIRC3 was deleted in 57 11q-deleted cases (83%) and in 4 cases (6%) the proximal deletion breakpoint resulted in partial deletion of BIRC3 that is predicted to cause protein truncation by removal of the RING domain. Additional deletion events not including ATM were identified in our series, but none of these encompassed BIRC3. In ATM deleted and non-deleted cases, the frequency of an ATM mutation of the retained allele was 38 and 21% respectively. In 11q-deleted cases with a wild-type ATM allele, three harboured both a deletion and mutation of BIRC3 (5% of BIRC3-deleted CLL). Mutations were located in either the CARD (frameshift mutation) or RING domains (STOP codon point mutation & 3bp indel) of the protein, and two are predicted to eliminate the C-terminal RING domain responsible for proteasomal degradation of MAP3K14. These three cases were also positive for trisomy 12, lacked an ATM mutation or TP53 abnormality, and two had mutations within the PEST domain of NOTCH1. Patients with deletion and mutation of BIRC3 were all treated (FC or F) within 39 months (mean TFS = 16 mths; range: 1–39mths) from diagnosis and after 9 years of follow up only one patient survived (mean OS = 83 mths). In conclusion, given the concordance between ATM and BIRC3 loss, and the low incidence of BIRC3 mutations, much larger studies would be required to evaluate whether disruption of BIRC3 has independent prognostic significance in an untreated cohort. Disclosures: No relevant conflicts of interest to declare.

Description: In CLL, over-expression of ZAP70 is well established as a prognostic biomarker for unfavourable disease course. However, technical challenges limit the application of flow cytometry for ZAP70 detection and measurement of ZAP70 mRNA expression is complicated by the need for T cell depletion. Having identified a correlation between DNA methylation of a 5' regulatory region of ZAP70 and expression of the gene, we and others have proposed the measurement of ZAP70 methylation as an alternative biomarker. Of note, recent work by Claus et al showed ZAP70 methylation to be prognostic for progression free survival (PFS) after therapy in two small, phase II, trial cohorts, treated with regimes incorporating Rituximab. However, as no comprehensive assessment of the prognostic value of ZAP70 methylation has been reported, we analysed ZAP70 methylation in 416 samples from patients entered into the phase III UK LRF CLL4 trial (comparing Chlorambucil(C) with Fludarabine alone (F) or in combination with Cyclophosphamide(FC)). The methylation of a single CpG site 334bp downstream of the transcription start site (C334) was interrogated in peripheral blood mononuclear cell DNA samples, using a quantitative bisulfite pyrosequencing assay. All samples were bisulfite converted, amplified and pyrosequenced in triplicate with a 6 point standard curve in each run. Standard curves for each of the 20 runs were linear (r2 〉 0.98) and sensitivity of pyrosequencing was ±5%. C334 methylation ranged from 0-100% and showed a significant inverse correlation with percentage of ZAP70 positive cells, IGHV identity to germline and percentage of CD38 positive cells (all p≤0.001). Methylation levels showed a trimodal, W-shaped, distribution with major modes at 5% and 95% and a minor mode at 50%. Based on this distribution, samples were classified as having Low (C334L66.6%) methylation (Figure 1).Figure 1Distribution of methylation levelsFigure 1. Distribution of methylation levels In univariate Kaplan Meier (KM) analysis, no significant difference was seen between C334L and C334Int groups in terms of PFS or OS (median PFS 1.68 versus 1.76 years, p=0.36 and median OS 4.71 years versus 5.41 years, p=0.18). However, C334H cases had a significantly longer PFS and OS than either group (median PFS=2.86 years, median OS=8.60 years, all p

Description: Most CLL is diagnosed with a low tumor burden with no indication for therapy. Biomarkers, such as unmutated IGHV genes, TP53 loss/mutation and raised β2M predict short time to first treatment and overall survival; however there remain patients with good risk biomarkers who nevertheless develop progressive disease. Advances in genomics and immunogenetics have lead to the discovery of new biomarkers and their integration with cytogenetic data refines outcome prediction. However these novel markers are predominantly found in IGHV unmutated cases (U-CLL). To identify novel genetic mechanisms that may contribute to progression, we have studied 13 patients (pts) presenting with cMBL (n=3) or Stage Binet A/Rai 0 disease and good risk markers: IGHV-mutated, excluding poor risk stereotypes (n=13), no 17p or 11q deletion by FISH, (n=13), sole del13q14 (n=8), low CD38 expression (n=13) who all developed lymphocytosis (n=13), between two untreated timepoints (TP1 & TP2), 10 of whom subsequently required treatment. Copy number analysis (SNP6), whole exome sequencing (WES; Agilent SureSelect & Illumina sequencing) of tumour-germline pairs and targeted deep sequencing (TDS; Haloplex, Agilent) of the the WES-identified variants and the 22 most frequently mutated genes in CLL, to a mean depth of 3681 fold, were performed at TP1 and TP2. TP1 was close to diagnosis (median of 1 yr, range 0.11-7.33) with a median time to TP2 of 4.5 yrs (0.2-8.9). In addition, TDS was undertaken at later time points in one patient described in point 4), who relapsed and ultimately transformed. Our analysis shows the following potential mechanisms: 1. Our germline WES data revealed 5 heterozygous missense/frameshift variants in 5 genes in 5 pts, also known to be targeted by somatic mutation in CLL (eg: FBXW7, POT1, SAMHD1. Fig1). 2. We then established the somatically-acquired mutation profile of each patient. We validated 72% (224/312) of the mutations discovered by WES using TDS and identified clinically relevant mutations earlier on in disease, supporting the hypothesis that sub-clonal mutations in genes in addition to TP53 may drive a progressive clinical course. At diagnosis (TP1) by WES/TDS, 5/13 pts had mutations in CLL driver genes (ATM, NOTCH1, SF3B1, TP53) and 2/13 pts had mutations in genes of undetermined clinical significance (CHD2, NFKBIE, ZMYM3). One patient was MYD88 mutated at TP1 and remains untreated after follow up of 12 yrs. In total, the following 9 genes (ATM, CHD2, DDX3X, MYD88, NOTCH1, NFKBIE, SF3B1, TP53 & ZMYM3) were mutated in 62% (8/13) pts at TP1. 3. Of the remaining 5/13 pts lacking a detectable mutation in any of the established CLL genes, we observed on average 7 mutations/patient in genes involved in cancer and each patient harboured one or more mutated genes with a role in haematological malignancy (eg. ITGA6, KLHL6, LTF, TNFAIP3). 4. One patient exhibited a remarkable temporal shift in copy number changes and mutations. At TP2, SNP6 analysis could not detect the del13q observed at TP1, and a clonal trisomy 12 had emerged, along with several mutations associated with progressive disease (BIRC3, IRF4, NOTCH1), that predominate in U-CLL. As a consequence we re-analysed the IGHV mutational status at TP2, and showed that rather than the IGHV3-48 with 92% germline identity identified at diagnosis, our patient exhibited an additional and dominant IGHV5-10-1*01 (100% identity) clone at TP2, 8 yrs after TP1. Additional analysis of intermediate samples detected the unmutated clone as far back as 4 yrs post diagnosis, and TDS analysis showed the NOTCH1 mutation was a minor subclone at diagnosis (0.06% VAF). Ultimately, this patient developed Richters syndrome with expansion of the NOTCH1 mutation (27% VAF). Retrospective sequential immunogenetic analysis of the other 12 cases yielded no other example of this phenomenon. In summary, IGHV-mutated cMBL/early stage CLL with a progressive outcome can be associated with, the presence of germline or subclonal gene mutations of known or putative importance in CLL, or the emergence of a IGHV-unmutated clone. Our data supports deep sequencing in the clinical setting for earlier detection of pathogenetic mutations and emerging immunogenetically distinct subclones in patients with early stage 'good risk' disease. Figure 1: Heatmap representation of the cohorts clinical features and DNA mutation. Patient 287 haboured the IGHV-unmutated clone at TP2-5. Figure 1:. Heatmap representation of the cohorts clinical features and DNA mutation. Patient 287 haboured the IGHV-unmutated clone at TP2-5. Disclosures No relevant conflicts of interest to declare.

Description: Splenic marginal zone lymphoma (SMZL) is an indolent B cell malignancy whose diagnosis is based on lymphocyte morphology, immunophenotype and marrow and/or splenic histology. Unlike other lymphomas, there is not a common chromosomal translocation specific for SMZL, and genetic prognostic factors are poorly defined. To investigate the pattern of genomic aberrations in SMZL, we applied comparative genomic hybridization to BAC microarrays (array CGH) to a well characterized series of 75 SMZL specimens. We applied two different 1 Mb-resolution BAC arrays: UCSF HumArray 3.2 and a novel array CGH platform developed at Univ. of Salamanca. These arrays allowed us to detect DNA copy number changes across the genome with high accuracy in 67 of 75 patient samples. Data were compared with our previous array CGH studies of 170 samples from different B-cell lymphoma subgroups. FISH studies for IGH, IGK and IGL translocations and 7q deletion were performed on tissue microarrays in 24 cases. Of the 67 samples, 19 (28%) showed a normal genomic profile. The median number of genomic aberrations per tumor was 2.2 (1.3 gains and 0.9 losses), which was lower than the rates detected in other lymphoma subgroups (diffuse large cell lymphoma, 6.4; mantle cell lymphoma, 6; follicular lymphoma, 4.5) and comparable to MALT lymphomas (2 abnormalities per tumor). SMZL cells showed a genomic pattern characterized by gain of chromosomes 3q24-q29 (18%), 6p (9%), 12q (9%), and 18q (4%) and loss of 7q32 (34%), 8p21-p23 (13%), 17p13 (10%) at P53 locus and 6q21-q27 (9%). Notably, no alterations of the P16/ARF (9p21) or MYC loci (8q24) were detected. Correlation of array CGH data with conventional cytogenetics, FISH and LOH studies revealed a high concordance. Detailed mapping of 7q deletions delineated a consensus region of loss of 3 Mb in 7q32. This 7q deletion was almost exclusive to SMZL, being observed in only 5 of 170 non-SMZL B-cell lymphomas (p=0.0000001). Four cases presented IG-translocation. Mutation of IGH was observed in 62% and correlated with a complex karyotype (61 vs. 13%; p=0,0008) whereas unmutated IGH correlated with the deletion of 7q (56 vs. 23%; p=0,01). Among the various genomic abnormalities, only the deletion of 8p or the presence of a complex karyotype correlated with inferior overall survival (OS) (median OS, 58 vs. 110 months, p=0,004; and 60 vs. 105 months, p=0,01; respectively). In summary, array CGH has defined a pattern of genomic aberrations in SMZL that differs from other B-cell lymphoma subgroups and that may predict overall survival. Because the deletion of 7q32 is the most distinctive genetic marker in SMZL, the identification of a putative tumor suppressor gene inactivated within the region of deletion seems mandatory.