Acute Myeloid Leukemia (AML) is an aggressive hematological malignancy caused by somatically acquired changes affecting a well-defined set of genes1. While rare high-risk variants affecting specific transcription factors account for a proportion of myelodysplastic syndrome (MDS) and AML associated with a family history, the contribution of other germline variants conferring low-intermediate risk has not yet been determined, partly because these are more difficult to identify from pedigree analysis. Here we use an Australian AML patient cohort to analyze rare, deleterious variants affecting genes involved in the rare recessive bone marrow failure syndrome Fanconi Anemia (FA). FA is caused by bi-allelic germline mutations in any of the 22 FANC genes (except for FANCB and FANCR which are X-linked and autosomal dominant), and is associated with profoundly increased risk of AML. The proteins encoded by the FANC genes participate in the removal of interstrand crosslinks (ICL) and the protection and resolution of stalled replication forks, an essential step for faithful DNA replication. Deficiency for these genes, combined with other mutations, results in pre-leukemia or leukemia in mouse models.

RBC-transfusion dependency (RBC-TD) is an independent prognostic factor for poor overall survival (OS) in the WHO classification-based prognostic scoring system (WPSS) for MDS patients. However, WPSS did not include cytopenia, whereas revised International Prognostic Scoring System (IPSS-R) did not include RBC-TD. Thus, neither of these prognostic scoring systems incorporates both cytopenia and RBC-TD. We aimed to test whether RBC-TD adds prognostic value to the IPSS-R. We analyzed MDS patients not treated with disease-modifying therapy, and enrolled in SA-MDS Registry (derivation cohort; n = 295) and Dusseldorf registry (Germany; validation cohort; n = 113) using time-dependent Cox proportional regression and serial landmark analyses. In the derivation cohort, RBC-TD patients had inferior OS compared to RBC transfusion-independent (RBC-TI) patients (P < 0.0001) at 6- (18 vs. 64 months), 12- (24 vs. 71 months), and 24-months (40 vs. 87 months). In a Cox proportional regression analysis, RBC-TD was an independent adverse prognostic marker in addition to age, sex, and IPSS-R variables (P < 0.0001). A prognostic index (PI) was derived using these Cox-proportional regression model variables. In the validation cohort, this PI classified patients into four prognostic groups with significantly different OS (P < 0.001) as in the derivation cohort. In conclusion, multivariate analysis by Cox proportional hazards regression and serial landmark analyses clearly demonstrates that development of RBC-TD at any time during the course of MDS is associated with poor OS, independent of IPSS-R. This study demonstrates that dynamic assessment of RBC-TD provides additional prognostic value to IPSS-R and should be included in treatment decision algorithms for MDS patients.

While there have been extensive studies to define the roles of recurrent somatic mutations in AML, the contribution of germline variants to AML initiation and progression is less well established. DNA repair disorders often predispose patients to developing myeloid malignancies. In particular, biallelic mutations affecting FANC genes cause the recessive heritable bone marrow failure syndrome Fanconi Anemia (FA), which is associated with >800-fold increased risk of progression to AML. A recent explosion of cancer predisposition studies has also revealed the importance of germlineFANC variants in elevated cancer risk (Cancer Treat Rev 2012; 38:89). To investigate the role of FANC gene variants in AML we have performed a case-control study, analyzing rare, deleterious somatic and germline variants for the 19 FANC genes in adult AML and healthy controls cohorts.

Introduction: Anemia is one of the commonest presenting features of MDS and approximately 30-40% of patients require regular RBC-transfusion. RBC-transfusion dependency (RBC-TD) is a poor-prognostic factor independent of revised International Prognostic Scoring System (IPSS-R) (Hiwase et al ASH 2014). Although RBC transfusion increases the risk of alloimmunization, there is limited literature characterizing this risk in MDS patients as compared to other hematological disorders (such as thalassemia).

S44 Context: The clinical implementation of next-generation sequencing has allowed for the quantitative detection of clinically significant somatic mutations in myelodysplastic syndromes (MDS). However, the clinical relevance of variant allele frequency (VAF) for the majority of mutations is unknown. Objective: To determine the importance of VAF in highly recurrent gene mutations for which genotype-phenotype relationships have been characterized in MDS and the impact of TP53 VAF on survival. Design: We profiled TP53 and up to 20 additional genes in our training set of 219 patients with MDS or secondary AML. Clinical variables and outcomes of these patients were characterized at the time of sample procurement. VAF was evaluated for genes with well described genotype-phenotype relationships with findings confirmed in a validation cohort. Fisher’s exact and Mann-Whitney’s tests were used for comparative analyses. Kaplan-Meier estimates were used to estimate overall survival and analyzed from the date of mutation identification. Results: When parsed by VAF, TP53 VAF strongly predicted for complex cytogenetics in both the training (P 1⁄4 .001) and validation set (P 40% had a median overall survival (OS) of 124 days versus an OS that was not reached in patients with VAF 40% was an independent covariate (HR, 1.61; P < .0001). For validation, TP53 VAF predicted survival in an independent cohort (HR, 4.94, P 1⁄4.01). Additionally, SRSF2 VAF predicted the presence (P 1⁄4 .003) and magnitude (P 1⁄4.004) of monocytosis, RUNX1 VAF with thrombocytopenia (P1⁄4.01), and SF3B1 with ringed sideroblasts (P < .0001). Conclusions: Allele burden influences phenotype penetrance across multiple genes. Notably, TP53 VAF improves prognostic precision compared to binary mutational analysis alone. Our study indicates that VAF should be incorporated in patient management and risk stratification in MDS.

Abstract Acute myeloid leukemia (AML), which is the most common acute leukemia in adults, is a particularly devastating disease that is universally fatal without therapy and has the longest hospital length of stay of any cancer. In addition there are a number of familial diseases described that have as a feature an increased susceptibility to AML. Examples of this are MDS/AML associated with germline mutations in the gene encoding the transcription factor GATA2, the ribosomopathies such as Diamond Blackfan Anemia, and Fanconi Anemia (FA) which is caused by bi-allelic mutations in any of the genes in FA complementation groups. Interestingly genes associated with predisposition to hematological malignancy, such as CEBPA, RUNX1 and GATA2 have also been described as targets of somatic mutation in AML. The application of next-generation whole-genome and whole-exome sequencing is greatly facilitating the dissection of the pathways altered during AML development. The discovery of new classes of mutations is increasing knowledge of the leukemogenesis process, improving disease classification and providing potential therapeutic targets. We have performed whole exome capture and next generation sequencing (NGS) of a series of 96 diagnostic AML samples. Analysis of the NGS data and comparison with genotyping performed using traditional methods, and a custom-designed Sequenom mutation panel, has confirmed the sensitivity and specificity of the NGS approach. Novel variants have been grouped into gene networks using a number of approaches, including pathway analysis, protein interaction databases, and known functional relationships. We have identified novel variants in a number of cancer susceptibility genes, and in particular we are characterising potential pathogenic mutations in a network of genes involved in the pathogenesis of Fanconi Anemia. Given that FA patients have a greatly increased predisposition to AML (at least 600-fold), we are performing functional assays to determine the significance for AML of somatic and germline variants in this network. Data will also be presented describing the clinical characteristics, mutation profile, gene expression signatures, and treatment outcomes for patients with mutations in this network. Citation Format: Anna L. Brown, James X. Gray, Paul Leo, Maung Kway Zeya, Mahmoud Bassal, Grant Engler, Sarah Bray, Brooke Gardiner, Mhairi Marshall, Ing Soo Tiong, Nik Cummings, Andrew Wei, Bik To, Ian Lewis, Alan D'Andrea, Thomas Gonda, Richard D'Andrea. Somatic mutation of cancer susceptibility genes in acute myeloid leukemia. [abstract]. In: Proceedings of the AACR Special Conference: Cancer Susceptibility and Cancer Susceptibility Syndromes; Jan 29-Feb 1, 2014; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(23 Suppl):Abstract nr 23. doi:10.1158/1538-7445.CANSUSC14-23

Introduction - AML is a complex group of malignancies, with heterogeneity in morphology, cytogenetics, molecular characteristics, aggressiveness and importantly, in its response to treatment and survival outcomes. Next generation sequencing by the Cancer Genome Atlas Research Network analysed 200 primary AML cases and identified 23 genes that display recurrent somatic mutations at varying frequency in AML ( NEJM 368(22):2059-2074). Defects in DNA repair are frequently identified in treatment-related AML and inherited mutations in genes of DNA repair pathways predispose patients to myeloid malignancies. For example, biallelic mutations in FANC genes, which cause the recessive heritable bone marrow failure syndrome Fanconi Anaemia (FA) are associated with high risk of progression to AML and other cancers (Kutler et al. Blood , 101:1249-1256), suggesting a potential involvement of FANC gene mutations in AML pathogenesis. Methods - In this study we present a two-stage approach to gene discovery in AML: initial unbiased whole genome sequence (WGS) and whole exome sequence (WES) analysis of tumour DNA from a cytogenetically normal AML case at diagnosis and relapse, and corresponding germ-line DNA (prepared from mesenchymal stromal cells). Potential oncogenic mutations and changes associated with disease progression were identified. WES of a further 96 diagnostic AML samples further defined recurrent mutations and allowed identification of affected functional groups and networks in AML. Results – WGS and WES were performed on diagnosis, non-haematopoietic and relapse samples from an index AML patient. Somatic SNVs and indels unique to the tumour samples include a number of variants in genes previously reported as recurrently somatically mutated in AML including FLT3, WT1 and IDH2. Somatic mutations in genes not previously associated with AML were also identified including a mutation in FANCD2 (p.S1412N) present in the index AML tumour DNA at diagnosis and at relapse. Variants in genes recurrently mutated at low frequency in AML can also be disease drivers, however separating such genes from the background level of mutation in AML requires analysis across multiple samples, and sequencing studies to determine recurrence and/or mutations in proteins involved in the same functional pathway or complex. STRING-db v9.05 (Franceschini et al . NAR , 2013(41), Database issue) was used to identify a larger network of proteins, including and associated with the FANC genes, involved in homologous recombination-mediated DNA repair. Known somatic mutations from other AML studies were mapped onto this network; as shown in [Figure 1][1] multiple genes in this extended network are affected by somatic mutation in AML suggesting a potential role in pathogenesis. Analysis of our WES data from diagnosis samples from a further 96 Australian AML cases identified an additional two somatic mutations in genes from the extended STRING-db v9.05 FANC network. In total we identified 18 mutations in the 16 classified FANC genes and 8 variants in the BLM complex as shown in [Figure 2][2]. Two of the germline FANC gene mutations, FANCM-Q13333fs and FANCD2-R926X, are known pathogenic mutations in FA. Patients with mutations in the 8 FANC genes of the core complex form a distinct subset from those with mutations in the other 8 FANC genes. 5 of the 8 patients with mutations in the BLM complex also form a separate group while BLM complex mutations are present in 2 patients that also have FANC mutations. For the two patients with acquired changes the allele frequency for these FANC mutations is greater than 25% suggesting an early origin in disease. Discussion. Our findings suggest that germline and somatic mutations affecting function of the FANC DNA repair pathway may be a recurrent abnormality in AML, potentially contributing to leukaemogenesis. FANC/BLM gene mutations frequently co-exist with mutations in DNMT3A and DNMT1; 46% of the patients with DNMT3A/DNMT1 mutations are also mutant for FANC or BLM complex genes representing significant over-representation (p = 0.021). Within the group of FANC and BLM patients there is also significant under-representation of FLT3-ITD mutations and mutations in N-RAS and K-RAS (p = 0.051), raising the possibility that defects in homologous DNA repair may favour cooperation with alternative signalling pathways. ![Figure 1][3] Figure 1 ![Figure 2][3] Figure 2 Disclosures No relevant conflicts of interest to declare. [1]: #F1 [2]: #F2 [3]: pending:yes

Abstract Background The International Prognostic Scoring System (IPSS) is commonly used for predicting the outcome of myelodysplastic syndrome (MDS) patients. Recently, a Revised IPSS (IPSS-R) has been developed to address the limitations of IPSS. IPSS-R identifies five different categories and stratifies patients better than IPSS. Although transfusion dependency is associated with inferior survival outcome, it has not been included in the risk stratification of IPSS-R mainly due to limited availability of transfusion data on patients used for deriving the IPSS-R. Aim To evaluate the impact of RBC transfusion on survival outcome in IPSS-R subgroups and assess the validity of IPSS-R in an independent cohort of patients. Materials and Methods To match the patient selection criteria used for generating the IPSS-R scoring system, primary MDS patients who were not treated with disease modifying agents or stem cell transplantation were included for this analysis. The impact of RBC transfusion on overall survival (OS) was assessed in IPSS-R subgroups. RBC transfusion dependency was defined as transfusion of at least 1 unit/8 weeks for at least 4 months. Results A total of 182 patients were included in this analysis. Their median age was 73 years (21 to 91 years) and 66% patients were male. 106 patients were in the Very Low or Low risk groups (termed ‘lower risk'). The median OS of IPSS-R Very Low, Low, Intermediate, High and Very High risk groups was 87.1, 63.9, 24.5, 17.2 and 7.8 months, respectively (Fig.1. p<0.0001), consistent with previously published results (Greenberg et al, Blood 2012). Of the 182 patients, 115 (63%) patients were RBC transfusion dependent. RBC transfusion dependency was more frequent in Very High (18/18, 100%), High (25/28, 89%) and Intermediate (21/31, 68%) risk groups as compared to lower risk IPSS-R groups: Low (35/67, 52%) and Very Low (17/39, 43%). The mean pre-transfusion Hb was 79.1 ±12.3 gm/L, and the trigger for transfusion was Hb ≤90, >90 to ≤100 and >100 gm/L in 83%, 11% and 6% of episodes, respectively. In a multivariate analysis, RBC-transfusion dependency (HR 3.18; P<0.0001) was associated with poor survival, independent of the IPSS-R category and age at diagnosis (Table 2). The median OS of transfusion-dependent patients (n=115) was significantly lower (23.8 vs. 117.8 months; p<0.0001) than that of transfusion-independent patients (n=67). As the majority of IPSS-R higher risk patients were transfusion dependent, we restricted further assessment to IPSS-R lower risk groups. The median OS between Low and Very Low risk group was not significantly different (87.1 vs Low 63.2 months; p=0.1), hence they were grouped together. The median OS of transfusion-dependent lower risk IPSS-R patients (n=52) was significantly shorter than that of transfusion-independent (n=54) patients (52.7 vs 122.5 months; p=0.001). Conclusions We have demonstrated that transfusion dependency is associated with inferior survival even in Very Low and Low risk IPSS-R group patients. This warrants further refinement of IPSS-R scoring system specifically for lower risk group patients. IPSS-R scoring system is validated in our independent cohort of patients. Disclosures: Hiwase: Novartis Australia: Research Funding; Celgene Australia: Research Funding.

We present the first case of simultaneous muscle-specific kinase antibody positive myaesthenia gravis and relapsing–remitting multiple sclerosis to be reported in the English literature along with the inherent diagnostic and treatment challenges. There may be an association between myaesthenia and central nervous system demyelination. We identified 72 previously published cases of myaesthenia with central nervous system demyelination. Of 19 cases of myaesthenia with relapsing–remitting multiple sclerosis, nine (47%) were acetylcholine receptor antibody negative, but there were no previously published cases with muscle-specific kinase antibody. Further research is required to clarify this association and optimal treatment in such cases.

WT1 levels may be a useful predictor of leukemia free survival (LFS) following treatment of acute myeloid leukemia (AML). We report a retrospective study in which levels of WT1 expression from patients with de novo AML were measured from bone marrow and peripheral blood at diagnosis, post-induction, post-consolidation and relapse. We demonstrate that higher levels of WT1 in peripheral blood at diagnosis are associated with poorer LFS independent of age and cytogenetic risk-group (n=85, p=0.028). When measured at post-consolidation, the presence of detectable WT1 is associated with poorer LFS in univariate analysis of both peripheral blood (p=0.024) and bone marrow (p=0.019). In a multivariate analysis including age and cytogenetic risk, the association remained significant for bone marrow (p=0.016) with a trend observed for peripheral blood (p=0.06). These findings have formed the basis for ongoing research.

Abstract 1690 Background – Acute Myeloid Leukaemia (AML) has a worldwide incidence of approximately 3.5 per 100,000 population per year with most cases occurring in adults. Survival at 12 months is less than 30% and at 5 years is less than 10% (NCI, 2010). AML is a complex disease that demonstrates marked heterogeneity morphologically, cytogenetically and molecularly. The largest cytogenetic subgroup of AML has normal cytogenetics and accounts for approximately 45% of de novo AML cases. Furthermore, the response to treatment and survival outcomes of cytogenetically normal AML (cn-AML) is remarkably heterogeneous. Aims – – To undertake whole-genome and exome sequence analysis of a cn-AML case at diagnosis and at relapse to identify subtle, potentially oncogenic mutations at the molecular level that may initiate AML and those that may be responsible for drug resistance or relapse. Specific Aims: i . –Comparison of the AML genomes at diagnosis, remission and relapse to identify genomic regions with changes in copy number and loss of heterozygosity, using single nucleotide polymorphism (SNP) microarray analysis. ii . Identify chromosomal rearrangements such as small insertions, duplications, inversions and deletions, (“indels”) and translocations that may contribute to AML initiation or relapse, using massively parallel, ultra-high throughput (“next generation”), paired-end sequencing of total genomic DNA at low depth. iii . Identify DNA point mutations potentially affecting protein function that may contribute to AML initiation or relapse by using next generation sequencing at high depth of the captured exomes (putative coding regions) of the AML and control genomes. Methods – The PwC (Price-Waterhouse-Coopers) Leukaemia and Lymphoma Tissue Bank, a joint initiative of the Australasian Leukaemia and Lymphoma Group and the Leukaemia Foundation, provided samples from an AML patient with normal cytogenetics and a blast count at diagnosis of 70% and at relapse of 85%. Genomic DNA extracted from autologous mesenchymal stem/stromal cells (MSCs) was used to represent non-leukaemic, germline, control DNA. Primary cell culture of MSCs from cryopreserved bone marrow aspirate cells (50 × 10 6 ) of our test patient at remission was achieved with standard tissue culture methods. High molecular weight DNA was extracted from the patient9s MSCs and marrow cells at diagnosis, remission and relapse samples. Sonication (Covaris) and libraries appropriate for paired-end high-throughput sequencing (Illumina Genome Analyzer II instrument) were prepared from gel purified DNA fragments approximately 200 bp in size. Results and Discussion – Primary cell culture of MSCs from bone marrow aspirates proved to be a robust source of germline genomic DNA with several advantages over skin. Firstly, tissue banked samples, will not include biopsies of normal skin. Furthermore, skin can be contaminated by circulating leukaemic cells, which is problematic with low-depth genomic sequencing. The homogenous immunophenotype of the cultured MSCs indicate their purity. Preliminary SNP microarray analysis identified a large region of uniparental disomy (copy number neutral loss of heterozygosity) involving most of chromosome 13q, which was not identified by standard cytogenetic analysis. Low-pass or shallow paired-end genomic DNA sequencing has generated the following outputs. MSC genome: 10.7 Gb (3.6X haploid genome coverage) resulting from 152 × 10 6 paired sequence reads, AML_diagnosis genome: 21.6 Gb (7.2X) resulting from 142 × 10 6 paired sequence reads, AML_relapse genome: 26.8 Gb (8.9X) resulting from 177 × 10 6 paired sequence reads. The capture baits representing the human exome span 26,225,870 bp (approximately 0.9% of haploid human genome). Exome capture libraries were prepared from the three sources of DNA (MSC, AML_diagnosis and AML_relapse) and sequenced using a single lane for each library. The outputs for all three were very similar, approximately 23 × 10 6 paired sequence reads, representing 1.3 Gb (49X haploid exome coverage). The reads for whole genome or exome capture were of high quality and more than 98% could be unambiguously aligned to the human reference genome in the correct orientation and interval distance. Here we will present MSC enrichment results and sequencing output results, quality control and preliminary analysis of genomic alterations and exonic mutations. Disclosures: No relevant conflicts of interest to declare.

Abstract 2639 This icon denotes an abstract that is clinically relevant.