ALBERT

Description: Introduction The STIL-TAL1 fusion is found in 16% cases of paediatric and adolescent T-ALL, making it one of the most common T-ALL subgroups. Our study considers this leukaemia subtype in the context of a complex ecosystem that is diverse, evolving and subject to selective pressures. We used single cell methods to understand the order of co-operating mutational events and the clonal evolution of mutations in genes that are re-iteratively targeted, such as PTEN. Methods Diagnostic DNA from five STIL-TAL1 positive T-ALL cases was exome sequenced using Agilent SureSelect Human all Exon kit plus Illumina paired end sequencing. Driver copy number alterations and NOTCH1/PTEN exon 7 mutation status had been identified in a previous study and candidate driver mutations for inclusion in single cell experiments were validated by sequencing or Q-PCR using custom assays. Where more than one mutation was present within the same exon of a candidate driver gene, cloning experiments were carried out to verify the independent mutation sequences. Material from xenograft transplants was available in three of the five cases to assess their clonal heterogeneity in the leukaemia initiating cell compartment. Single cell multiplex Q-PCR was used to examine the single cell genetics of the pre-defined mutation events. Briefly, single cells were sorted and lysed prior to multiplex specific (DNA) target amplification and Q-PCR using the 96.96 dynamic microfluidic array and the BioMark HD (Fluidigm, UK). Copy number assays for the 1p33 deletion and custom assays for the patient specific STIL-TAL1 fusion breakpoints were used to confirm that the 1p33 deletion leading to this gene fusion was a clonal event. Results The only aberrant events common to all five samples were CKDN2A copy number loss and the 1p33 deletion that results in the STIL-TAL1 fusion. Exome sequencing revealed further mutations in known T-ALL drivers including NOTCH1, PTEN and PHF6 as well as candidate driver mutations in FREM2, PIK3CD, RPL14, BMPR1A and CDH18. Both NOTCH1 and PTEN demonstrated re-iterative inactivation and this was investigated in detail for PTEN. Case 1 had multiple PTEN exon 7 mutations and sub-clonal copy number loss. Case 2 had parallel frameshift mutations in PTEN exons 5 and 7. Case 3 contained an exon 8 mutation and multiple PTEN exon 7 mutations. In this case the three most frequent PTEN exon 7 indels were validated and tracked in a single cell multiplex Q-PCR experiment. This revealed a branching sub-clonal genetic architecture (see figure 1) in which all malignant cells at the proposed apex of the branching architecture harboured the STIL-TAL1 fusion and CDKN2A deletion with copy number losses of 4p, 6q and FREM2 and PTEN mutations occurring as sub-clonal events. PTEN indels 2 and 3 were found co-localised in the same sub-clone. Preliminary analysis of the paired mouse xenograft bone marrow did not detect PTEN exon 7 indels 1 – 3 in 84 single cells. However, bulk Sanger Sequencing analysis did identify the PTEN exon 8 mutation in the mouse. Ongoing work is in progress to determine whether single cells of the xenograft carry alternative PTEN exon 7 mutations detected in the diagnostic sample exome data and to characterise in which diagnostic sub-clone the PTEN exon 8 mutation resides. Conclusions This study demonstrates how exome sequencing and single cell multiplex Q-PCR can be used as complementary tools to understand the sub-clonal complexity of STIL-TAL1 T-ALL. PTEN inactivation is sub-clonal by single cell analysis, demonstrating the parallel evolution of multiple independent PTEN inactivated sub-clones, highlighting PTEN inactivation as a key event in this T-ALL subgroup. In a wider cohort of 20 patients collected by our group at least 50% had PTEN inactivation as assessed by sequencing of exon 7 and copy number data alone. Results indicate a strong evolutionary pressure selecting for mutational events that result in inactivation of the PTEN-PI3Kinase pathway. These events occur via multiple mechanisms, including copy number loss and truncating mutations, which are not limited to the known T-ALL hotspot in exon 7. Current work is focussing on using a similar approach to examine the clonal evolution of NOTCH1 mutations in STIL-TAL1 T-ALL samples in diagnostic and xenograft samples of cases 4 and 5. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.

Description: Abstract 2521 Purpose: The outcome for Teenagers and Young Adults (TYA) with T-cell Acute Lymphoblastic Leukaemia (T-ALL) has improved significantly using paediatric based chemotherapy protocols. Event free survival figures from the recently closed UKALL2003 highlight this observation. It remains poorly understood however why their outlook remains inferior to children aged 1 – 15 years whilst receiving identical treatment. Possible explanations for this phenomenon are distinct tumour biology, host factors and treatment adherence. The primary aim of this study is to characterise the tumour associated genetic events leading to TYA T-ALL and establish whether T-ALL in Teenagers and Young Adults is a distinct disease entity. The secondary aims are to identify new prognostic markers and potential targets to optimise therapy. Methods: DNA and RNA were extracted from 60 TYA T-ALL (15 – 25 years) historic samples. The DNA was derived from blasts at diagnosis or relapse, and matched with germ line DNA where available. The DNA was examined at high resolution for Copy Number Alterations (CNAs) and Loss of Heterozygosity (LOH) using the Affymetrix SNP6.0 platform and analysed by CNAG, dCHIP and Partek. The most frequent CNAs were confirmed by MLPA (P-383-T-ALL). In addition the DNA was screened for mutations in NOTCH1/FBXW7/PTEN using conventional Sanger sequencing and for mutations in NRAS, KRAS, CBL, FLT3 and SHP2 using denaturing high performance liquid chromatography. In cases without CDKN2A gene deletion we initially performed CDKN2A gene promoter methylation by SABiosciences EpiTect Methyl qPCR Arrays, followed up by Methylation Specific PCR (MSP). Where available, RNA was extracted from viable cells and the quality confirmed by Bioanalyser. Next the RNA was reverse transcribed to cDNA and subjected to real-time PCR gene expression analysis for CDKN2A and T-ALL related oncogenes. Results: The most frequent CNA was deletion of CDKN2A in 72.7% of patient samples (91.6% homozygous deletions). Other frequent CNAs were loss of MLLT3 (31%), STIL (27%), PTEN (19%), LEF1 (15%) and gain of MYB (12%). These CNAs were confirmed by MLPA. The STIL-TAL fusion occurred in 27% of patients. The only recurrent region of Copy Number Neutral LOH encompassed 9p24.3-p13.3, including the gene CDKN2A (27%). NOTCH1, FBXW7 and PTEN mutations occurred at expected frequencies. Rare mutations were identified in NRAS and CBL, and these mutations represented known variants. Bisulphite modified DNA was subjected to CDKN2A gene promoter methylation analysis. All samples examined were negative by Epitect Methyl Array, including the positive control cell line Raji. Promoter methylation was positive by MSP for the Raji cell line and some TYA T-ALL cases without CDKN2A gene deletion. Conclusions: This data identifies the frequency of recurrent mutations in T-ALL of Teenagers and Young Adults but reveals no striking differences with those observed in T-ALL in younger children. We are in the process of analysing additional cases to complete a total of 75 patients, including whole exome sequencing of a subset of cases. This sample size will allow us to detect genomic aberrations with a 5% incidence within this cohort (〉90% probability) and to determine any prognostic significance. The most frequently observed aberration is CDKN2A gene deletion. In some cases without CDKN2A deletion we identified CDKN2A gene promoter methylation, offering an alternative mechanism of CDKN2A inactivation. To be able to appreciate potential cooperation between genetic abnormalities, we are correlating aberrant expression of the oncogenes TAL1, LYL1, LMO1, LMO2, TLX1, TLX3, NKX2, ERG and MEF2C with the genomic abnormalities identified. Disclosures: Gribben: Celgene: Honoraria; Roche: Honoraria; Merck: Honoraria; Mundipharma: Honoraria; Pharmacyclics: Honoraria.

Description: The t(12;21) translocation that generates the ETV6-RUNX1 (TEL-AML1) fusion gene, is the most common chromosomal rearrangement in childhood cancer and is exclusively associated with B-cell precursor acute lymphoblastic leukemia (BCP-ALL). The translocation arises in utero and is necessary but insufficient for the development of leukemia. Single-nucleotide polymorphism array analysis of ETV6-RUNX1 patient samples has identified multiple additional genetic alterations; however, the role of these lesions in leukemogenesis remains undetermined. Moreover, murine models of ETV6-RUNX1 ALL that faithfully recapitulate the human disease are lacking. To identify novel genes that cooperate with ETV6-RUNX1 in leukemogenesis, we generated a mouse model that uses the endogenous Etv6 locus to coexpress the Etv6-RUNX1 fusion and Sleeping Beauty transposase. An insertional mutagenesis screen was performed by intercrossing these mice with those carrying a Sleeping Beauty transposon array. In contrast to previous models, a substantial proportion (20%) of the offspring developed BCP-ALL. Isolation of the transposon insertion sites identified genes known to be associated with BCP-ALL, including Ebf1 and Epor, in addition to other novel candidates. This is the first mouse model of ETV6-RUNX1 to develop BCP-ALL and provides important insight into the cooperating genetic alterations in ETV6-RUNX1 leukemia.

Description: B-cell precursor childhood acute lymphoblastic leukemia with ETV6-RUNX1 (TEL-AML1) fusion has an overall good prognosis, but relapses occur, usually after cessation of treatment and occasionally many years later. We have investigated the clonal origins of relapse by comparing the profiles of genomewide copy number alterations at presentation in 21 patients with those in matched relapse (12-119 months). We identified, in total, 159 copy number alterations at presentation and 231 at relapse (excluding Ig/TCR). Deletions of CDKN2A/B or CCNC (6q16.2-3) or both increased from 38% at presentation to 76% in relapse, suggesting that cell-cycle deregulation contributed to emergence of relapse. A novel observation was recurrent gain of chromosome 16 (2 patients at presentation, 4 at relapse) and deletion of plasmocytoma variant translocation 1 in 3 patients. The data indicate that, irrespective of time to relapse, the relapse clone was derived from either a major or minor clone at presentation. Backtracking analysis by FISH identified a minor subclone at diagnosis whose genotype matched that observed in relapse ∼ 10 years later. These data indicate subclonal diversity at diagnosis, providing a variable basis for intraclonal origins of relapse and extended periods (years) of dormancy, possibly by quiescence, for stem cells in ETV6-RUNX1+ acute lymphoblastic leukemia.

Description: Abstract LBA-5 Intra-clonal, mutational complexity is a hallmark feature of cancer and provides the substrate for sub-clonal selection, progression and therapeutic resistance. There is limited insight however into detailed sub-clonal, genetic architecture in cancers and their propagating ’stem’ cells. Childhood acute lymphoblastic leukaemia (ALL) genotypes are characterized by chimeric fusion genes (or hyperdiploidy) coupled with recurrent, copy number alterations (CNA), primarily in genes regulating cell cycle or differentiation. For ETV6-RUNX1-positive cases of B precursor ALL, the temporal sequence of events is known with the fusion gene usually arising as a pre-natal, initiating event. Recurrent CNA, presumed to be functional or ’driver’ mutations, are secondary to gene fusion and probably post-natal in origin. We have interrogated clonal architecture by multiplexed FISH analysis of single cells using probes labelled with three or four distinct fluorochromes. We pre-selected diagnostic samples from 30 ETV6-RUNX1-positive cases that we found to have deletion of ETV6 and CDKN2A or PAX5 or combinations of these. The application of multiple FISH probes allowed us to score all cells (200 per patient sample) for up to six genetic lesions: ETV6-RUNX1 fusion (and multiple copies of the fusion), deletion of the unrearranged ETV6 allele, extra copies of chromosome 21q (via RUNX1 signals), and one or two copy deletions of PAX5 and CDKN2A. The genetic classification of individual cells using this method allowed a designation of sub-clones and the assembly of putative ancestral, or evolutionary, trees. Although sub-clones were previously known to exist at diagnosis of ALL, the extent of sub-clonal diversity revealed by this study was unanticipated and very marked. Sub-clones, of varying sizes (1–90% of total cells), numbered 3 to 14 per case. This must be an under-estimate of genetic complexity as only selected mutations were included in the screen. The common CNA (ETV6, CDKN2A and PAX5 deletions, chromosome 21q gains) arose independently and recurrently in sub-clones and in no preferential order. Matched relapse and diagnosis pairs were available on 5 patients. In all of these, the clones at relapse could be matched to individual sub-clones at diagnosis, with relapse originating from a major or minor subclone at diagnosis. Importantly, the relapse clone often diversified further, reiterating the evolution pathway of the primary clones. These data indicate that sub-clonal diversification does not arise via linear, clonal succession but rather has a complex, branching architecture reminiscent of Charles Darwin’s 1837 evolutionary speciation model. The ancestral trees so revealed in ALL are single time point snapshots and therefore disguise temporal or sequential dynamics. We show that sub-clonal architecture changes in the months leading up to a diagnosis of ALL and is re-ordered by treatment and subsequent relapse. An important prediction derived from this pattern of sub-clonal architecture is that leukaemia propagating or ’stem’ cells in ALL (and other cancers) should themselves be genetically diverse. Preliminary transplantation experiments in NOD/SCID/IL2Rgamma(null) mice show that this is the case, since multiple, genetically distinct sub-clones regenerated in vivo (Figure1). The complexity of genetic architecture in ALL has substantial implications for the cancer stem cell concept and for the efficacy of generic or targeted treatments. Disclosures: No relevant conflicts of interest to declare.

Description: The timing and developmental sequence of events for BCR-ABL1+ acute lymphoblastic leukemia (ALL), usually associated with IKAROS (IKZF1) deletions, are unknown. We assessed the status of BCR-ABL1 and IKZF1 genes in 2 pairs of monozygotic twins, one pair concordant, the other discordant for Philadelphia chromosome positive (Ph+) ALL. The twin pair concordant for ALL shared identical BCR-ABL1 genomic sequence indicative of monoclonal, in utero origin. One twin had IKZF1 deletion and died after transplantation. The other twin had hyperdiploidy, no IKZF1 deletion, and is still in remission 8 years after transplantation. In the twin pair discordant for ALL, neonatal blood spots from both twins harbored the same clonotypic BCR-ABL1 sequence. Low level BCR-ABL1+ cells were present in the healthy co-twin but lacked the IKZF1 deletion present in the other twin's leukemic cells. The twin with ALL relapsed and died after transplantation. The co-twin remains healthy and leukemia free. These data show that in childhood Ph+ ALL, BCR-ABL1 gene fusion can be a prenatal and possibly initiating genetic event. In the absence of additional, secondary changes, the leukemic clone remains clinically silent. IKZF1 is a secondary and probable postnatal mutation in these cases, and as a recurrent but alternative copy number change is associated with poor prognosis.

Description: Background Germline mutations in the N-terminal of CCAAT/enhancer binding protein α (CEBPA) are a feature of autosomal dominant AML. Despite the strong penetrance of these mutations, the age of disease onset varies considerably and is usually precipitated by acquiring a C-terminal mutation. Although biallelic CEBPA-mutations in sporadic AML are associated with favorable clinical outcomes, little is known about long-term survival or the secondary molecular events linked with familial cases. Aims We sought to establish the long term clinical outcomes in familial CEBPA-mutated AML and to examine the patterns of secondary mutations associated with leukemic transformation. Methods and Results Disease specific and follow up information was collected in 16 affected patients, from 7 pedigrees, published between 2004 and 2011. In 94% of patients (n=15), at least 3 years follow up was achieved. All pedigrees had a germline N-terminal CEBPA mutation and 17 of 18 documented disease episodes had an acquired C-terminal mutation. The age at AML diagnosis varied from 2-39 years (median 24.5 yrs) with a single asymptomatic carrier detected (now 23 yrs). With the exception of 1 patient diagnosed in 1963, all cases received combination chemotherapy at diagnosis. Additional consolidation comprised autologous stem cell transplantation (SCT, n=3) and allogeneic SCT in 1 patient failing to achieve CR post induction therapy. Ten patients had at least 1 further disease episode, the first recurrence presenting after a median of 2.1 years (range 0.5-14 yrs), 5 continued in CR and 1 patient was lost to follow up. In 3 out of 4 patients, where CEBPA was screened at recurrence, the acquired C-terminal mutations differed from diagnosis, signifying new episodes of AML. The median overall survival (OS) for the entire cohort was 20 years (1.1-46 yrs, n=16) and 17.3 years for patients with multiple disease episodes, reflecting durable responses to second line therapies. To identify potential co-operating mutations in CEBPA pedigrees, whole exome sequencing (WES) was performed in 7 tumor samples from 5 patients across 3 pedigrees, all with the germline mutation p.P23fs (Figure 1). All tumor DNA samples were sequenced with matched remission or skin DNA as a germline control. The number of acquired mutations in familial tumors was similar to sporadic disease, with 10-22 (median=14) non-synonymous tier 1 mutations per sample. In addition to the acquired C-terminal CEBPA mutation, these included established AML loci such as EZH2, TET2, WT1, GATA2, NRAS, CSF3R and the recently identified cohesin complex gene, SMC3. Of note, FLT3-ITD, NPM1 and DNMT3A mutations were absent in all tumors. A minimum of 2 established mutations were identified in each tumor and, at present, we can only speculate on which additional mutations are ‘driver' or ‘passenger' events. Reflecting findings in sporadic AML, biallelic CEBPA and GATA2 mutations co-occurred in both siblings from Pedigree 1 and were subsequently identified by Sanger sequencing in the child III.2 (Figure 1). All 3 patients continue in long term remission following chemotherapy. We were able to trace the clonal evolution in patient I.2 (Pedigree 3) by WES of 3 consecutive tumor samples which arose over a 17 year period. At diagnosis (Dx) the patient received induction and consolidation chemotherapy and remained disease free for 14 years. The second disease episode (R1) was treated with chemotherapy followed by autologous SCT and the third presentation (R2) was chemo-refractory. Tumor DNA from R2 was clonally related to Dx, sharing 7 identical mutations, including the original C-terminal CEBPA deletion. In contrast, R1 appeared molecularly distinct from Dx and R2, most likely representing a new clone which was subsequently eradicated with treatment. Conclusion This is the first report of long term clinical outcomes in familial CEBPA-mutated AML. Although many patients experienced disease recurrence, our extended follow up showed that OS remained favorable despite multiple episodes of disease. Assessment of C-terminal CEBPA mutations provided a unique insight into the recurrence of AML, with some patients appearing to develop completely new leukemic episodes. Although the penetrance of germline mutations is high, healthy carriers and late onset disease are noted, emphasizing the need for clinical vigilance and screening of all related potential SCT donors. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 1348 Philadelphia-positive (Ph+) acute lymphoblastic leukaemia (ALL), characterised by the BCR-ABL1 fusion gene, occurs in approximately 30% of adult and 5% of childhood ALL and is associated with a poor prognosis. It is considered a single clinical entity with identifiable and recurrent copy number alterations (CNA); notably deletions of the lymphoid transcriptional regulator IKAROS (encoded by IKZF1), PAX5, and CDKN2A/B that are presumed to cooperate with BCR–ABL 1 in lymphoid leukaemogenesis. In particular, IKZF1 deletions are present in 80% of BCR-ABL1 positive ALL cases, and have been implicated as an independent indicator of poor prognosis in childhood ALL. Our previous studies of twin pairs either concordant or discordant for BCR-ABL1+ ALL indicate that the fusion gene is a first hit that occurs prenatally. However, the order and sequence of acquisition of CNA is unknown. We recently reported a complex sub-clonal genetic architecture for leukaemic blasts and leukaemia-propagating (‘stem’) cells in childhood ETV6-RUNX1-positive ALL (Anderson et al., Nature 469: 356–361, 2011). In the present study, we aimed to determine whether similar sub-clonal genetic diversity occurs in BCR-ABL1+ ALL. We carried out five colour FISH to diagnostic blast cells from eight BCR-ABL1 positive cases with differentially-labelled probes for BCR, ABL1, IKZF1, CDKN2A and PAX5. In a subset of cases we also performed Affymetrix single nucleotide polymorphism (SNP 6.0) arrays to determine the specific boundaries of deletions. Four out of the eight cases screened had concurrent IKZF1, PAX5 and CDKN2A deletions. In one case the order of acquisition of these deletions was uninformative, with 97% of cells exhibiting a single FISH pattern (BCR-ABL1+ with monoallelic deletions of all three genes). In the second case, a linear clonal progression was observed with IKZF1 deleted first, PAX5 second and CDKN2A third. In the two remaining cases a branching sub-clonal pattern was observed. In one of these monoallelic IKZF1, CDKN2A and PAX5 deletions all arose independently in different sub-clones; i.e. IKZF1 was deleted first in one subclone, CDKN2A first in another and PAX5 first in a third sub-clone. In the final case we also studied matched diagnosis and relapse samples. Here, SNP array analysis revealed different deletions in all three genes at diagnosis and relapse. Genomic fusion breakpoint analysis revealed an identical BCR-ABL1 genomic sequence at diagnosis and relapse, confirming the same clonal origin of leukaemia. The different deletion boundaries in IKZF1, PAX5 and CDKN2A permitted us to design specific FISH probes to distinguish between ‘diagnostic’ and ‘relapse’ deletions and to track their evolution. The predominant clone at relapse was not a direct evolutionary product of any of the major clones found at diagnosis. The dominant sub-clone at diagnosis was BCR-ABL1+, with a large 9p deletion (encompassing PAX5 and CDKN2A) and a focal CDKN2A deletion, all sub-clonal to a focal IKZF1 deletion. At relapse, the dominant sub-clone had acquired a different IKZF1 deletion, which was sub-clonal to two different (focal, biallelic) deletions of CDKN2A and a different monoallelic PAX5 deletion. The large 9p deletion was not present at relapse. These results indicate the existence of a pre-leukaemic BCR-ABL1 fusion gene positive clone that has given rise to at least two sub-clones, each with different IKZF1, PAX5 and CDKN2A deletions, that have evolved independently. These data indicate that the sub-clonal architecture in this poor prognosis subtype of ALL is genetically diverse, and that key ‘driver’ CNA can arise independently and in no preferential order. Disclosures: No relevant conflicts of interest to declare.