ALBERT

Description: Chronic lymphocytic leukaemia (CLL) is the most common form of adult leukaemia worldwide. Patients display a highly variable clinical course, with some requiring immediate therapeutic intervention while others can remain untreated for years. We have previously reported that DNA methylation of the homeobox A4 (HOXA4) promoter can serve as part of a three gene prognostic signature with CD38 and BTG4 to predict time to first treatment (TTT) in Stage A patients. HOXA4 encodes a transcription factor that is expressed in haematopoietic progenitor cells and is involved in embryonic development and B-cell differentiation, and its aberrant epigenetic regulation has been identified in multiple forms of leukaemia. In this study we have sought to elucidate the role of HOXA4in the progression of CLL and determine the functional consequences of its expression. We analysed DNA methylation of the HOXA4 promoter by pyrosequencing in a heterogeneous cohort of 163 CLL patients (median age: 70; median follow-up: 10 years), of whom 60% were Binet Stage A, 16% Stage B and 24% Stage C. Data was collected regarding treatment history, TTT and overall survival, as well as cytogenetic abnormalities and IGVH mutation status. HOXA4 methylation increased with disease progression and was significantly higher in Stage C patients (median 74%) than those with Stage A (62%; p = 0.03) and Stage B disease (65%; p 〈 0.05). HOXA4 methylation was positively correlated with IGVH sequence homology (r = 0.34, p 〈 0.0001) and negatively associated with TTT among patients who have started chemotherapy (p = 0.04) and with overall survival (p = 0.04). No associations were observed between HOXA4 methylation and 11q, 13q or 17p deletions, or TP53 and ATMmutations. To investigate the role of HOXA4 in the evolution of the disease, we analysed samples taken at multiple timepoints from 42 patients, of whom 29 were undergoing treatment and 13 remained untreated. HOXA4methylation significantly increased in patients undergoing treatment (p = 0.01), but did not differ in untreated patients (p = 0.19). We hypothesised that silencing of HOXA4 may be selected for during treatment due to its expression conferring increased sensitivity to chemotherapy. Using a lentiviral system, we observed that re-expression of HOXA4increased drug sensitivity in a malignant differentiated B-cell line (Raji). Significantly higher apoptosis was identified after treatment with 3 μM and 10 μM fludarabine (both p 〈 0.001) and 1 μM and 10 μM ibrutinib (p 〈 0.01 and p 〈 0.001), but not 1 μM and 10 μM idelalisib. To confirm the translational relevance our observations, we overexpressed HOXA4 in primary CLL cells derived from four patients and confirmed increased apoptosis in response to 3 μM and 10 μM fludarabine treatment in comparison to control cells (p = 0.02 and p 〈 0.01). Further work is underway in primary CLL cells to elucidate the pathways under the control of HOXA4 that may confer this drug sensitivity. Our ongoing work may indicate that HOXA4 is also implicated in the progression of CLL through directing malignant cells to the protective bone marrow niche, thereby further reducing sensitivity to antimetabolites. In cell lines HOXA4 up-regulates the expression of RGS2 and RGS16, which are negative regulators of the CXCR4-CXCL12 signalling axis, and we have identified selection for biallelic HOXA4methylation in primary acute lymphoblastic leukaemia cells following engraftment in mice (median in primary cells: 80%; engrafted cells: 92%; p 〈 0.0001). To determine the origins of HOXA4 dysregulation during the course of the disease, we analysed prospective blood samples from the European Prospective Investigation into Cancer and Nutrition (EPIC) from 20 individuals diagnosed with CLL up to 17 years after blood draw (median: 7 years) and 20 age-matched controls who remained free of cancer. We observed that HOXA4methylation was significantly higher among future CLL patients (median: 49% vs 42%; p = 0.01) and was inversely correlated with time to diagnosis, but did not reach statistical significance (r = -0.39, p = 0.09). Together, our findings suggest that silencing of the HOXA4 gene is an early event in CLL which is selected for during the course of disease through reduced sensitivity to chemotherapeutic agents. Our ongoing work will identify downstream targets that may be implicated in conferring sensitivity, and which may serve as biomarkers to predict prognosis and inform treatment strategies. Disclosures Junge: AstraZeneca: Other: Salary, Research Funding.

Description: Loss of heterozygosity (LOH) is detectable in many forms of malignancy, including leukaemia, using techniques such as microsatellite analysis and comparative genomic hybridisation. However, these techniques are laborious and require the use of relatively large amounts of DNA if the whole genome is to be examined. Here we describe the use of oligonucleotide microarrays to characterise single nucleotide polymorphisms (SNPs) in lymphoblasts isolated from children with acute lymphoblastic leukaemia for the pan-genomic mapping of LOH with a resolution of 100–200kb. Results were compared with DNA obtained during remission and on relapse. Abnormalities were seen in 8 of 10 cases. The two cases with no abnormalities and one case which showed identical changes affecting whole chromosomes at relapse and presentation remain in remission 1–9 years following retreatment. The 7 cases which showed LOH not affecting entire chromosomes died following relapse, suggesting that partial LOH may be associated with a poor prognosis. In 4 cases LOH was only detectable at relapse suggesting that progressive LOH may be a cause of disease progression and/or drug resistance. This was supported by detailed analysis of one case in which LOH involving the glucocorticoid receptor (GR) was associated with mutation of the remaining allele. In cell line models the loss of a functional GR is associated with profound resistance to steroids. The most frequent abnormality detected in this series involved chromosome 9p. In each of the four cases where this was observed LOH included the INK4 locus. In three of the four cases INK4 loss was only observed at relapse (see figure), suggesting that this abnormality may be commonly associated with treatment failure, supporting previous reports that 9p abnormailities are associated with a poor prognosis. One case was reported as showing monosomy 20 as the sole cytogenetic aberration but LOH analysis identified 9p LOH and loss of 20q, with retention of heterozygocity for 20p. These findings strongly implicate unbalanced translocation der(9)t(9;20),-20 as described by Clark et al (Leukaemia, 2000, 14:241). Our observations demonstrate that SNP array analysis is a powerful new tool for the analysis of allelic imbalance and unbalanced translocations in leukaemic blasts.

Description: Introduction Acute myeloid leukaemia is a heterogenous disease with variable response to chemotherapy. In order to prognosticate at an individual level numerous cytogenetic and molecular markers may have to be taken into account. Most publications in AML relate to clinical trials and outcomes in this context. We aimed to study outcome in a population-based cohort in the era of molecular genetic testing. Methods All patients, aged 19 and over, diagnosed with AML between 2007-2011, throughout the north east of England (population 3.1 million) were identified. This was done by searching weekly multidisciplinary team meeting minutes across the three haematology teams in the region and triangulating these data with cytogenetic and molecular genetic data. Only patients aged 19-60 years (inclusive) at diagnosis are reported. All biopsy specimens were subject to central pathology review. Results A total of 344 patients were identified and 150 were aged 19-60. Nineteen patients with acute promyelocytic leukaemia (APL) were excluded. Twelve patients were excluded due to missing data; thus 119 non-APL were analysed: 66 women and 53 men. All patients were considered suitable for intensive therapy and 58 (49%) were included in a national AML trial. Ninety eight out of 119 patients (82%) achieved a complete remission (CR); 79 patients entered CR post cycle 1. 21 patients (17%) did not enter a CR (four died before treatment could commence, nine died during induction, six were refractory and palliated and 2 became aplastic and died before remission status could be ascertained). Thirty-nine patients (40%) subsequently relapsed after achieving CR, 19 of these were successfully re-induced and all but one had an allograft in CR2. Eleven patients failed re-induction and were subsequently palliated and one received an allograft for refractory disease. With a median follow up of 1699 days, the median overall survival (OS) for the population was 603 days. Cytogenetics was a strong predictor of survival with median OS (days) being 225, 508 and not reached (NR) for poor (n=29), standard (n=75) and good (n=15) cytogenetic risk groups respectively (p

Description: Chromosomal abnormalities are increasingly used to risk stratify adults with acute lymphoblastic leukemia. Published data describing the age-specific incidence of chromosomal abnormalities and their prognostic relevance are largely derived from clinical trials. Trials frequently have age restrictions and low recruitment rates. Thus we investigated these factors in a population-based cohort of 349 patients diagnosed during the course of 19 years in the northern part of England. The incidence of most chromosomal abnormalities varied significantly with age. The incidence of t(9;22)(q34;q11) increased in each successive decade, up to 24% among 40- to 49-year-old subjects. Thereafter the incidence reached a plateau. t(4;11)(q21;q23) and t(1;19)(q23;p13) were a rare occurrence among patients older than 60 years of age. In contrast, the frequency of t(8;14)(q24;q32) and t(14;18)(q32;q21) increased with age. High hyperdiploidy occurred in 13% of patients younger than 20 years of age but in only 5% of older patients. The incidence of low hypodiploidy/near-triploidy and complex karyotype increased with age from 4% (15-29 years) to 16% (≥ 60 years). Overall survival varied significantly by age and cytogenetics. Older patients and those with t(9;22), t(4;11), low hypodiploidy/near-triploidy, or complex karyotype had a significantly inferior outcome. These population-based results demonstrate the cytogenetic heterogeneity of adult acute lymphoblastic leukemia. These data will inform the delivery of routine clinical services and the design of new age-focused clinical trials.

Description: Introduction. Current treatment protocols for chronic lymphocytic leukemia (CLL), including FCR (fludarabine, cyclophosphamide and rituximab), have dramatically improved response rates. However, since none of these therapies is curative, continued preclinical studies on innovative therapeutic strategies are warranted. In particular, the identification of new agents that interfere with the survival of CLL cells by promoting their apoptosis is one approach to improve therapeutic outcome. Taking into account that p53 deletions and/or mutations in CLL are restricted to 10%-15% of patients at diagnosis, we have investigated the efficacy of the second-generation MDM2 antagonist RG7388, in primary CLL samples, examining its ability to disrupt the p53-MDM2 interaction, resulting in expression of p53 transcriptional targets that trigger apoptosis. Methods. Heparinized peripheral blood samples were obtained from CLL patients (n=42) with informed consent, in accordance with institutional guidelines and the Declaration of Helsinki. Mononuclear cells were purified by Lymphoprep density-gradient centrifugation. Cells were resuspended in RPMI 1640 medium supplemented with 10% FCS and 1% penicillin/streptomycin. Cells were treated with increasing concentrations of RG7388 and incubated at 37°C, 5% CO2. Ex vivo cytotoxicity was assessed in all samples 48 hours after treatment using an XTT assay. A pilot time-course experiment identified 6 hours after treatment as the most informative time point for gene expression analysis. Therefore, in 8 samples cells were treated for 6 hours and RNA was extracted using standard methods. qRT-PCR on a panel of nine p53 transcriptional targets (CDKN1A, MDM2, PUMA, BAX, FAS, FDXR, ZMAT3, TP53INP1, DR5) was performed using SYBR Green chemistry. The functional integrity of p53 was assessed by western blot following treatment with MDM2 antagonists in 20 samples. Apoptosis was assessed using a Caspase 3/7 luminescent assay. Results. Treatment with RG7388 significantly decreased cell viability (LC50 〈 1µM) in 28/42 CLL samples (mean LC50 = 0.40 ± 0.05 µM). As expected, the 8 CLL samples that displayed non functional p53 on western blot and/or mutated TP53 by sequencing were more drug-resistant (mean LC50 = 5.22 ± 1.13 µM) than those sensitive responders (LC50 〈 1µM) with a functional p53 (mean LC50 = 0.39 ± 0.05 µM, n=12). Interestingly, among p53 functional samples, we identified a small subset that showed intermediate response (1 µM 10 µM, n=3) to RG7388 treatment. RG7388 significantly upregulated the mRNA expression of MDM2 and of p53 transcriptional targets involved in the intrinsic pathway of apoptosis (PUMA and BAX), and the extrinsic pathway of apoptosis (death receptor 5 (DR5)) (Figure 1A, n=5 samples). Interestingly, only a slight upregulation of CDKN1A was observed and upregulation of pro-apoptotic genes dominated. Overall, in p53 functional CLL samples, exposure to 3 µM RG7388 for 6 hours led to a 7.8-fold increase in MDM2, a 7-fold increase in PUMA, a 4.9-fold increase in DR5, a 4.5-fold increase in BAX, and a 3.3-fold increase in CDKN1A (Figure 1A). Conversely, no significant upregulation of p53 target genes was seen in p53 non-functionalCLL samples (Figure 1B, n=3 samples). Treatment with RG7388 led to a concentration-dependent increase in caspase 3/7 activity in p53 functional CLL samples (Figure 1C). No significant difference in caspase 3/7 activity was observed in p53 non-functional samples after RG7388 treatment (Figure 1D). Interestingly, when we analysed samples taken two years apart from a fludarabine- and chlorambucil-resistant CLL patient, we found that treatment with RG7388 decreased cell viability, induced p53 accumulation and upregulation of p53-target genes, suggesting that inhibiting the p53-MDM2 interaction might be a promising treatment strategy in chemo-resistant CLL patients that show a functional p53. Conclusion. Taken together, our data indicate that RG7388 induces a characteristic dominantly pro-apoptotic gene expression profile of p53-target genes in primary p53-functional CLL samples. RG7388 showed a consequent potent apoptotic effect on CLL cells, which was dependent on the presence of a functional p53. Identification of patients in whom functional p53 activation drives apoptosis may translate into improved outcomes for patients treated with non-genotoxic MDM2 antagonists. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.

Description: Inactivation of the tumor suppressor gene, CDKN2A, can occur by deletion, methylation, or mutation. We assessed the principal mode of inactivation in childhood acute lymphoblastic leukemia (ALL) and frequency in biologically relevant subgroups. Mutation or methylation was rare, whereas genomic deletion occurred in 21% of B-cell precursor ALL and 50% of T-ALL patients. Single nucleotide polymorphism arrays revealed copy number neutral (CNN) loss of heterozygosity (LOH) in 8% of patients. Array-based comparative genomic hybridization demonstrated that the mean size of deletions was 14.8 Mb and biallelic deletions composed a large and small deletion (mean sizes, 23.3 Mb and 1.4 Mb). Among 86 patients, only 2 small deletions were below the resolution of detection by fluorescence in situ hybridization. Patients with high hyperdiploidy, ETV6-RUNX1, or 11q23/MLL rearrangements had low rates of deletion (11%, 15%, 13%), whereas patients with t(9;22), t(1;19), TLX3, or TLX1 rearrangements had higher frequencies (61%, 42%, 78%, and 89%). In conclusion, CDKN2A deletion is a significant secondary abnormality in childhood ALL strongly correlated with phenotype and genotype. The variation in the incidence of CDKN2A deletions by cytogenetic subgroup may explain its inconsistent association with outcome. CNN LOH without apparent CDKN2A inactivation suggests the presence of other relevant genes in this region.

Description: The regulation of hematopoietic lineage fate and commitment is fundamental to normal and malignant hematopoiesis. Switches between lymphoid and myeloid lineages in leukemia are rare and associated with poor clinical outcome, but potentially very informative regarding the regulation of hematopoietic lineage commitment. In contrast to therapy-related acute leukemia (AL) after a first primary leukemia, lineage-switch ALs arise from a common pre-leukemic or leukemic clone and share a founder mutation, most often rearrangement of MLL at 11q23. The majority of switches are from acute lymphoblastic leukemia (ALL) to acute myeloid leukemia (AML); however, conversions from myeloid to lymphoid and even oscillations between the two lineages have been observed, although the molecular mechanisms underlying lineage switch have not yet been identified. Here we describe a male patient who presented at 9 months of age with a t(4;11)-positive B-ALL and was subsequently treated according to the Interfant06 protocol. He achieved complete remission, but relapsed at the age of 4 years with a t(4;11)-positive AML. He underwent allogeneic BM transplantation and has remained in remission 13 months. Sanger sequencing revealed identical translocation breakpoints in the ALL and AML samples, demonstrating a lymphoid to myeloid lineage switch with a common pre-leukemic or leukaemic cell of origin for both ALs. Interestingly, whereas the AML shows no V(D)J rearrangements, we found incomplete rearrangements in the ALL cells indicating a ProB cell origin. In line with this observation, B-ALL cells expressed 6-fold and 120-fold higher levels of PAX5 and EBF1, respectively, compared to AML blast cells. Microsatellite instability measurements argued against a strong therapy-associated impairment of DNA mismatch repair in the AML. The translocation t(4;11) is the most frequently found chromosomal rearrangement in infant leukaemia and is almost exclusively associated with ALL at presentation, suggesting a strong instructive potential towards the lymphoid cell fate. However, the occurrence of lineage switch in t(4;11) AL demonstrates that this instruction can be overcome by as yet unknown mechanisms. Exome sequencing identified 16 and 98 novel somatic variants in the diagnostic ALL sample (0.23 mutations per Mb) and AML (1.4 mutations per Mb), respectively, of which 10 were shared. Of the total novel somatic mutations, there were 1 and 12 non-synonymous alterations in the B-ALL and AML, respectively, of which one was shared. RNA sequencing confirmed that all 12 genes with non-synonymous mutations were expressed in AML blast cells, and all belonged to the top 25% of expressed genes in both the AML and B-ALL. Genes carrying non-synonymous somatic AML-specific mutations include CHD4 (12p13, NuRD helicase, chromatin maintenance, DNA repair, lineage fidelity, part of MLL complex), NCOA2 (8q13, also known at TIF2 and part of the MOZ/TIF2 fusion gene, cofactor of nuclear receptors including VDR and NR3C1; control of myeloid differentiation, putative tumour suppressor), CEP164 (11q23, centrosome protein involved in microtubule organization, DNA damage response and chromosome segregation) and PPP1R7 (2q37, regulatory subunit of protein phosphatase 1, control of mitosis, regulator of AURKB). All amino acid residues predicted to be mutated are conserved between several species. Each of the identified mutations is located in functionally relevant regions and may, thus, interfere with protein function. Notably, exome and RNAseq showed that all 12 of the AML-specific non-synonymous mutations were found in at least 40% of the reads covering the corresponding genomic positions (the sample analysed constituted 80% blast material), thus suggesting heterozygosity for each mutation and that all mutations are common to the major leukemic clone. Taken together, these data suggest that the B-ALL and AML share a common ancestral pre-leukemic or leukemic cell of origin. Whilst the B-ALL revealed few novel somatic mutations, the change in lineage is associated with the acquisition of a substantial number of novel mutations indicating significant clonal evolution preceding the emergence of myeloid neoplasia. These data identify candidate mutations/genes which may overcome lineage instruction by a leukemic master regulator such as MLL/AF4 and which may therefore play an essential role in the control of hematopoietic lineage fate. Disclosures: No relevant conflicts of interest to declare.

Description: Genetic alterations including chromosomal translocation, promoter hypermethylation, somatic mutation and gene deletion are believed to play a key role in the leukemogenic process in childhood acute lymphoblastic leukemia (ALL). The p16INK4a (CDKN2A/MTS1/p16/INK4a) gene located on chromosome 9p21 is a tumor suppressor gene whose product can block cell division during the G1/S phase of the cell cycle. Inactivation of p16INK4a in ALL can occur by deletion, promoter hypermethylation or somatic mutation. However, published reports are inconsistent in terms of both incidence and route of p16INK4a inactivation suggesting that a detailed analysis of all possible modes of inactivation in a large cohort is essential to clarify the status of this gene in leukemogenesis. In this study, we report the findings of a comprehensive analysis of p16INK4a in 115 DNA samples with childhood ALL (86 cases at presentation and 29 cases at relapse) in which a combination of techniques including, fluorescence in situ hybridization (FISH), mapping arrays, denaturing high performance liquid chromatography (dHPLC) and methylation specific-PCR (MSP) were used to assess the mode of inactivation of this gene. Data from a genome-wide screening in 86 presentation cases and 20 of 29 relapse cases using Affymetrix Mapping 10K and/or 50K single nucleotide polymorphism (SNP) microarray technique showed loss of heterozygosity (LOH) at the p16INK4a locus in 21% (22/106) of cases (14 at presentation and 8 at relapse), 14 (8 at presentation and 6 at relapse) with an associated loss of copy number and 8 (6 at presentation and 2 at relapse) with a normal copy number, indicative of acquired isodisomy (AID). FISH analysis on 19 of the 22 confirmed that those cases with LOH and copy number loss had either p16INK4a homozygous (n=6) or hemizygous (n=6) deletion and those with LOH associated with AID (n=7) retained 2 copies. Mutation and methylation analyses were performed on those cases identified to have one p16INK4a allele or retention of both alleles. Partial methylation of p16INK4a was found in only 1 case. Mutational screening by dHPLC of the coding region revealed a somatic mutation, H83Y, in a subpopulation of leukemic blasts in one patient at relapse. Three common SNPs were identified including A148T in exon 2 and 500C〉G and 540 C〉T in the 3′ UTR. These data show that mutation and hypermethylation of p16INK4a are rare events in childhood ALL but that homozygous and hemizygous deletion is relatively common. The loss of only one p16INK4a allele in this latter group, without evidence for mutation or hypermethylation of the remaining one suggests that p16INK4a may be haploinsufficient in ALL. The finding that LOH on 9p locus is common but in nearly 40% of these cases is associated with AID with intact p16INK4a, suggests the existence of another tumor suppressor gene or oncogene in this region, which may have importance in leukemogenesis.