ALBERT

Description: The last three authors contributed equally to this work. Introduction: Allogeneic hematopoietic stem cell transplantation (HSCT) may improve long-term survival of patients with acute lymphoblastic leukemia (ALL) at high risk of relapse. However, ovarian failure is observed in 70 to 90% of patients who received myeloablative HSCT. Autotransplantation of cryopreserved ovarian cortex harvested before gonadotoxic treatments has been shown to re-establish the menstrual cycles and lead to the birth of healthy children (Donnez J. N Engl J Med, 2018). In patients treated for acute leukemia, concern has been expressed about the risk of leukemia recurrence after the ovarian autotransplantation due to the potential presence of residual leukemic cells in the ovarian tissue (Dolmans MM. Blood, 2010). To date, there were only few studies evaluating the presence of leukemic cells in cryopreserved ovarian tissue, and in most cases ovarian tissues were harvested at diagnosis of ALL or early after onset of chemotherapy. The present study prospectively investigated, in the context of fertility preservation, the presence of leukemic cells in cryopreserved ovarian samples harvested before allogeneic HSCT in patients in complete remission (CR) of ALL. Patients and Methods: From October 2015 to May 2018, all female patients with ALL who had a leukemia specific marker and underwent ovarian cryopreservation before allogeneic HSCT as part of preservation fertility program in 3 centers were included in the study. Consents were obtained from the guardians and/or age appropriate patients. The specific local ethical committees approved the study. Ovariectomy was performed after patients achieved CR, mostly in the weeks preceding HSCT. Ovarian cortex was separated into several fragments and cryopreserved as previously described (Poirot C. Human Reprod, 2002). When more than 14 cortical fragments were obtained, one was dedicated to molecular analysis, as well as the ovarian medulla. MRD quantification was performed by quantitative PCR of clonal rearrangements of immunoglobulin/T-cell receptor genes (Ig/TCR) or oncogenic fusion genes, according to EuroMRD and European Against Cancer (EAC) guidelines. MRD in ovarian samples were compared to MRD in bone marrow (BM) sample obtained at the same time point. Results: Fourteen patients were included in the study: 12 with B-cell precursor ALL, 1 with T-cell ALL and 1 with mixed phenotype acute leukemia. MRD marker was Ig/TCR rearrangement in 12 cases, M-BCR-ABL transcript in 1 case and genomic MLL-AF4 in 1 case. Median age at transplant was 18.2 years (range 1.1 - 35.2 years). Criteria for HSCT were poor early response to treatment (N=8), Philadelphia positive ALL (N=1), or previous relapse (N=5). With a median follow-up after HSCT of 8.1 months (range 0.8 - 29.8 months), all patients are disease-free. At the time of ovariectomy, 7/14 (50%) patients had undetectable MRD in ovarian samples, 6 had low positivity below 10-4 and one had positive MRD higher than 10-4. Unexpectedly of the 7 patients with positive ovarian MRD, 4 were undetectable in BM. One of the 7 patients with undetectable MRD in ovarian samples was positive in bone marrow. Concordant results were observed between the cortex and medulla samples in 10 of the 11 cases where both could be tested. Conclusions: To our knowledge, this series is the largest cohort evaluating MRD in ovarian samples from patients with ALL in complete remission after full chemotherapy regimen. We detected low levels of residual leukemic cells in ovarian tissues in half of the patients, some of them having no detectable MRD in BM at the time of ovariectomy, suggesting that leukemic cells could preferentially persist in ovary. Our results warrant further analyses of an extended cohort and longer post-transplant follow-up to better assess the potential presence of leukemic cells in ovarian samples and evaluate the impact of ovarian MRD status on post HSCT relapse. Still, autotransplantation of cryopreserved ovarian cortex could be discussed after allogeneic HSCT in patients with undetectable ovarian MRD. Table 1. MRD evaluation in bone marrow and ovarian samples. BM, bone marrow; BCP-ALL, B-cell precursor ALL; Ph-positive ALL, ALL with Philadelphia chromosome; NA: not available. Undetectable MRD means negative result obtained with a sensitivity of 10-4 or 10-5. BCR-ABL1 MRD results are expressed as the BCR-ABL1/ABL1 transcripts ratio. Disclosures No relevant conflicts of interest to declare.

Description: The C-Myb transcription factor is essential for hematopoiesis, including in the T-cell lineage. The C-Myb locus is a common site of retroviral insertional mutagenesis, however no recurrent genomic involvement has been reported in human malignancies. Here, we identified 2 types of genomic alterations involving the C-MYB locus at 6q23 in human T-cell acute leukemia (T-ALL). First, we found a reciprocal translocation, t(6;7)(q23;q34), that juxtaposed the TCRB and C-MYB loci (n = 6 cases). Second, a genome-wide copy-number analysis by array-based comparative genomic hybridization (array-CGH) identified short somatic duplications that include C-MYB (MYBdup, n = 13 cases of 84 T-ALL, 15%). Expression analysis, including allele-specific approaches, showed stronger C-MYB expression in the MYB-rearranged cases compared with other T-ALLs, and a dramatically skewed C-MYB allele expression in the TCRB-MYB cases, which suggests that a translocation-driven deregulated expression may overcome a cellular attempt to down-regulate C-MYB. Strikingly, profiling of the T-ALLs by clinical, genomic, and large-scale gene expression analyses shows that the TCRB-MYB translocation defines a new T-ALL subtype associated with a very young age for T-cell leukemia (median, 2.2 years) and with a proliferation/mitosis expression signature. By contrast, the MYBdup alteration was associated with the previously defined T-ALL subtypes.

Description: TLX1 is a homeodomain transcription factor generally associated with a favorable outcome in T-cell acute lymphoblastic leukemia (T-ALL). However, the molecular mechanisms of TLX1 deregulation remain unclear and various transcript levels in the absence of 10q24 abnormalities have been reported. A reproducible and accurate delineation of TLX1+ T-ALL will be necessary for proper therapeutic stratification. We have studied 264 unselected T-ALLs (171 adults and 93 children) and show that T-ALLs expressing high levels of TLX1 (n = 35, 13%), defined as a real-time quantitative polymerase chain reaction (RQ-PCR) level of TLX1 greater than 1.00 ABL, form a homogeneous oncogenic group, based on their uniform stage of maturation arrest and oncogenetic and transcriptional profiles. Furthermore, TLX1-high T-ALLs harbor molecular TLX1 locus abnormalities in the majority (31/33), a proportion largely underestimated by standard karyotypic screening. T-ALLs expressing TLX1 at lower levels (n = 57, 22%) do not share these characteristics. Prognostic analysis within the adult LALA94 and GRAALL03 prospective protocols demonstrate a better event-free survival (P = .035) and a marked trend for longer overall survival (P = .059) for TLX1-high T-ALLs, while the expression of lower levels of TLX1 does not impact on prognosis. We propose that TLX1+ T-ALLs be defined as cases expressing TLX1/ABL ratios greater than 1 and/or demonstrating TLX1 rearrangement. Therapeutic modification should be considered for those patients.

Description: We have identified a new recurrent chromosomal translocation, targeting the major homeobox gene cluster HOXA in human T-cell acute lymphoblastic leukemia (T-ALL). Four cases were characterized using a combination of FISH, Southern blot, breakpoint region sequencing, and a large scale expression analysis of a series of T-ALL. Specific RQ-PCR analysis of the HOXA1 to HOXA13 transcripts showed that the whole HOXA gene cluster expression was dramatically deregulated in the HOXA-rearranged cases, and also in the MLL and CALM-AF10-related T-ALL, strongly suggesting that HOXA genes are oncogenic in these types of leukemia. The HOXA-rearranged cases were included in a general portrait of T-ALL based on large scale expression analysis, showing that a new homogeneous T-ALL subgroup is defined by this chromosomal rearrangement. Moreover, patterns of gene expression associated to the distinct T-ALL oncogenic subgroups were compared with gene expression in normal human thymic sub-populations (11 purified sub-populations). Inappropriate use or perturbation of some specific molecular networks involved in thymic differentiation could be detected in the T-ALL cells. Also, we found that abnormal, frequently ectopic, expression of at least one developmental gene, including HOXA, TLX1/HOX11, TLX3/HOX11L2 and a few more, could be identified in most of the T-ALL cases. Our data strongly support the view that the abnormal expression of developmental genes, including the prototypical major homeobox genes HOXA in some cases, is critical in T-ALL oncogenesis.

Description: D-type cyclins are key regulators of progression through G1 phase of the cell cycle. Strong expression of at least one of the three D cyclins is common in human cancers. However, while the cyclin D1 and D3 genes (CCND1 and CCND3) are recurrently involved in genomic rearrangements, especially in mantle-cell lymphoma and multiple myeloma, no clear involvement of the cyclin D2 gene (CCND2) has been reported to date in human malignancies. In T-cell acute lymphoblastic leukemia (T-ALL), the T-cell receptor genes TCRA/D and TCRB are frequently involved in chromosomal rearrangements. In order to identify new genomic rearrangements and oncogenes in human T-ALL, we performed an interphasic FISH screening of T-ALL cases using TCR flanking probes. Using this approach, we identified two new chromosomal translocations: t(7;12)(q34;p13) and t(12;14)(p13;q11), involving the TCRB and TCRA/D loci, respectively. Molecular analysis of the breakpoint sequences demonstrated the involvement of the CCND2 locus at 12p13. Expression analysis using RQ-PCR and immunoblotting demonstrated dramatic cyclin D2 overexpression in the translocated cases (n=3) compared to other T-ALLs (total, n=86), whereas other genes located near the translocation breakpoints were not deregulated on microarray analysis. To further evaluate expression in T-ALL with respect to normal T-cell differentiation, we analyzed CCND2 expression in purified subpopulations from normal human thymus. CCND2 levels were downregulated through progression from the early stages of normal human T-cell differentiation and transition to beta-selection. In the most immature T-ALLs, a moderate CCND2 expression was observed, consistent with their differentiation stage, while low expression was found in other T-ALL. By contrast, the massive and sustained expression in the CCND2-rearranged T-ALL cases strongly suggested an oncogenic role due to the TCR translocation. T-ALL oncogenesis is a multi-step process; we here found that the TCR-CCND2 translocations were associated with other oncogene expression (TAL1, HOXA, or TLX3/HOX11L2), NOTCH1 activating mutations, and/or CDKN2A/p16/ARF deletion, showing that cyclin D2 dysregulation could contribute to multi-event oncogenesis in various T-ALL groups. In conclusion, this report is the first clear evidence of a direct involvement of cyclin D2 in human cancer due to recurrent somatic genetic alterations. This reinforces the view that the strong expression of cyclin D2 which is detected in various types of cancer can contribute to oncogenesis, and points to cyclin D2 as a potential target for therapy in these tumors.

Description: Over the last years, numerous studies have dissected the mutational landscape of T-cell acute lymphoblastic leukemia (T-ALL) resulting in the identification of numerous oncogenes and tumor suppressors implicated in T-cell transformation. However, most genetic abnormalities found in cancer are located in intergenic regions, whose role in cancer development, if any, remains poorly understood. Here we hypothesized that recurrent cancer-associated intergenic mutations, amplifications and deletions may implicate strong transcriptional regulatory sequences responsible for the activation of key oncogenic factors in the pathogenesis of T-ALL. To address this question, we first used ChIPseq analysis to map the genomic landscape of enhancer sequences controlled by NOTCH1, a critical T-ALL oncogene activated by mutations in over 60% of human T-ALLs. In addition, we performed high resolution aCGH analysis of somatic chromosomal amplifications and deletions in a comprehensive series of 160 T-ALL samples. These analyses revealed recurrent focal duplications at chromosome 8q24 in 8/160 (5%) T-ALL cases in an area devoid of protein-coding genes containing a prominent 1 kb NOTCH1-binding peak. Strikingly, this putative oncogenic element is located +1,427 kb 3’ from the MYC locus and chromatin configuration 3C analysis demonstrated its direct association with the MYC proximal promoter. Multispecies DNA sequence alignment revealed remarkable conservation of this region in mammals, birds and reptiles and local ChIP analysis revealed bona fide active enhancer features including P300 occupancy and high levels of H3K4me1 with low levels of H3K4me3. In addition, detailed analysis of epigenetic marks across 64 hematopoietic and non-hematopoietic cell lines and tissues revealed that his regulatory element is located within a major superenhancer specifically active in T-cells. Based on these results, we proposed that this regulatory sequence, hereby named N-Me for NOTCH-bound MYCenhancer, could function as an important regulatory element driving the activation of MYC downstream of NOTCH1 in T-ALL. Consistently, luciferase reporter assays showed strong, orientation-independent and NOTCH-dependent activation of reporter constructs containing N-Me in association with a -2.5 kb MYC proximal promoter in JURKAT T-ALL cells. Next and to formally test the functional relevance of this enhancer in T-cell development and transformation we generated and characterized N-Me knockout and conditional knockout mice. N-Me null animals were viable and showed a marked and selective reduction in thymus size and cellularity as their only developmental alteration. Detailed immunophenotypic analysis of N-Me knockout thymocytes demonstrated accumulation of double negative 3 (DN3) T cells and marked reductions in double positive and CD4+ and CD8+ single positive cells. Mechanistically, this phenotype was associated with a marked reduction in Myc expression in DN3, DN4 and ISP cells. Moreover, transplantation of N-Me knockout bone marrow hematopoietic progenitors demonstrated that this phenotype is cell autonomous, and can be rescued upon retroviral expression of MYC. Next, we analyzed the role of N-Me in the induction of NOTCH-driven leukemias by transplanting mice with N-Me wild-type and knockout hematopoietic progenitors infected with retroviruses expressing a mutant constitutively active form of NOTCH1 (ΔE-NOTCH1). In this context, mice transplanted with ΔE-NOTCH1-infected N-Me wild type cells developed T-ALL with 100% penetrance 6 weeks after transplant. In contrast, mice transplanted with ΔE-NOTCH1-N-Me knockout cells showed complete resistance to NOTCH1-induced transformation remaining 100% leukemia-free at 15 weeks post-transplant. In addition, secondary deletion of N-Me in established NOTCH1 induced leukemias from tamoxifen-inducible N-Me conditional knockout mice (Rosa26TMCre N-Meflox/flox) induced profound antileukemic effects with extended survival and almost complete suppression of leukemia initiating cell activity. Altogether, these results identify N-Me as the first long range oncogenic enhancer directly implicated in the pathogenesis of human leukemia. Disclosures No relevant conflicts of interest to declare.

Description: Abstract 121 The genetic landscape of B-cell precursor acute lymphoblastic leukemia (BCP-ALL) in children above 10 years and adolescents remains poorly defined. Specifically, more than half of these patients have none of the cytogenetic abnormalities that define oncogenic subtypes and underlie risk stratification. To uncover new genetic abnormalities in these unassigned cases, we studied 85 BCP-ALL from patients aged 10 to 17 diagnosed at St-Louis hospital (Paris, France), for which the main classifying genetic lesions were assessed (i.e. high hyperdiploidy, t(12;21)/ETV6-RUNX1, t(1;19)/TCF3-PBX1, t(9;22)/BCR-ABL1, iAMP21, MLL translocations, low hypodiploidy, and near haploidy). Fifty of these BCP-ALL presented no classifying genetic lesions. Paired leukemic and remission samples could be analysed by high density array-CGH (Agilent 1M arrays) in 17 of these unassigned cases. We focused on acquired, focal, and recurrent copy-number abnormalities. A mono-allelic intragenic deletion of the ETS-related Gene (ERG) was found in 3 cases. ERG belongs to the ETS family of transcription factors and is implicated in chromosomal translocations associated with several cancer types including acute myeloid leukemia. The possibility of a cryptic unbalanced translocation was ruled out in the 3 cases by FISH analysis. The deletions encompassed exons 3 to 7, or 3 to 9, and the breakpoints were tightly clustered. Based on the breakpoint sequences we designed a PCR assay that allowed us to screen ERG intragenic deletions in the whole cohort. ERG deletion was identified in 9 additional cases, none of them having any of the known classifying genetic lesions, bringing up to 25% (12 out of 50) the frequency of ERG deletion in unassigned BCP-ALL of children older than 10. These results suggested that ERG deletion characterized a novel oncogenic subtype of BCP-ALL. Of note, these results were consistent with independent data of Harvey et al. (2010) that reported ERG deletions in a distinct gene-expression cluster. To confirm and extend these findings in the whole population of paediatric BCP-ALL, we used our breakpoint-specific PCR assay to screen ERG deletions in an independent cohort of 822 unselected patients aged 1 to 17, enrolled in the EORTC 58951 trial. ERG deletion was identified in 31/822 (3.7%) patients. Again, none of them had another known classifying genetic lesion, confirming that ERG deletion characterizes a distinct oncogenic subtype. Patients with ERG deletion were significantly older compared to other patients (median 7.0 vs 4.0, p=0.002), but they had similar white blood counts at diagnosis. They had a favourable outcome, with a 8-year event free survival (EFS) of 82.4% and overall survival (OS) of 96.0%, which is similar to EFS of 83.4% and OS of 91.6% obtained for patients having no very high risk initial features (i.e. no t(9;22)/BCR-ABL1, MLL rearrangement or haploidy/low hypodiploidy). IKZF1 deletion is a cooperative genetic lesion that has been recently shown to be associated with a poor outcome in BCP-ALL. Remarkably, the incidence of IKZF1 deletions in patients with ERG deletion was significantly higher than in other BCR-ABL1-negative patients, especially when considering the IKZF1 intragenic deletion Δ4-7 (10/31, 32.3% vs 34/744, 4.6%, P

Description: Background: MDS and AML are mostly found in elderly patients. However, even in this population there is increasing evidence of predisposing genetic conditions, which have been underdiagnosed so far. Identifying inherited predisposition to myeloid disorders can be crucial especially in the context of hematopoietic stem cell transplantation (HSCT). Germline mutations in the DEAD/H-box helicase gene DDX41 have been identified in families with multiple cases of MDS or AML but also in sporadic cases. We aimed to analyze the prevalence and clinical features of DDX41-related myeloid malignancies within an unselected cohort of pts diagnosed with MDS or AML (MDS/AML). Methods Between March 2017 and June 2018, mutation screening was performed in 842 consecutive pts with a diagnosis of MDS/AML in a single center at Hôpital Saint-Louis, Paris. DNA was obtained from bone marrow or peripheral blood. Targeted sequencing of all exons of a panel of 80 genes recurrently mutated in myeloid malignancies was performed using custom capture-based library preparation (Agilent SureSelect) and Illumina sequencing. Sanger sequencing was performed on selected pts' cultured skin fibroblasts to check for the putative germline origin of the variants. Results We identified a DDX41 gene variant in 36 unrelated pts (4% of 842). We focused on the 32 pts having at least one DDX41 variant with a variant allele frequency (VAF) ranging from 40 to 60% highly suggestive of a germline origin, which was subsequently confirmed in all available cases (N=7). Sixteen variants were classified as pathogenic or likely pathogenic based on major predicted changes in protein sequence while the 16 others were missense variants of unknown significance (VUS), which scored deleterious in most algorithms (Figure 1A). An additional, likely somatic DDX41 mutation (VAF 〈 40%) was present in 18 of 32 pts (56%). Overall, 22 pts could be unambiguously considered as having a DDX41-related malignancy based on the presence of a major disturbing mutation and/or a second DDX41 mutation, while 10 pts had a single VUS. Twenty-six variants were newly described, including a recurrent one, G173R found in 5 pts, all having a second DDX41 mutation. Median age of the 32 patients was 70 years (35-88). Only 4 pts (12%) had a familial history of hematologic disorders. According to revised WHO classification, 4 (12.5%) had MDS-MLD, 8 MDS-EB (25%), 12 AML (37.5%), 6 MDS/MPN (18.7%), one 5q syndrome and one aplastic anemia. Strikingly, 15/32 (47%) pts had a history of cytopenia several years before blastic evolution and the 5 pts with G173R presented with hypoplastic MDS or initially isolated cytopenias, suggesting a specific functional effect of this mutation. Karyotype was normal in 16 pts (44%), complex in one, 12 pts had an isolated abnormality, and three had cytogenetic failure. Additional driver mutations were identified in most (27/32,84%) pts (Figure 1B), but we noticed that they were less frequent and at lower VAF in pts having both germline and somatic DDX41 mutations as compared to pts with a single variant (median 1.5 vs 3 mutations, median VAF 7% vs 29.5%, p