ALBERT

Description: Abstract 1717 Background: Aberrant DNA methylation of tumor suppressor genes is a common event in myelodysplastic syndromes (MDS) and acute myeloid leukemias (AML). Therefore hypomethylating agents like azacytidine (AZA) and decitabine seem to be a good therapeutic approach for the treatment of these diseases. Clinical experience and recent published data have demonstrated that AZA is effective for the treatment of MDS and AML patients. However, the prognostic impact of the aberrant hypermethylation on response and outcome to AZA treatment remains to be determined. For this reason the influence of the methylation status of a selected set of tumor suppressor genes on the overall survival and clinical response in MDS and AML patients, prior to treatment with AZA was studied. Patients and Methods: A total of 78 patients with MDS or AML who had been treated with AZA were evaluated. Among these, 25 were excluded because they had received less than 4 cycles, AZA was used after allogeneic stem cell transplantation, response was not assessable, or there was not enough quality DNA available. So finally, the study was focused in 53 patients: 36 MDS and 17 AML. Most of the AML included in the study had low blast count (20–30%). Responses were assessed according to the IWG MDS criteria in accordance to Fenaux et.al, and IWG AML criteria following European LeukemiaNet recommendations. DNA methylation status was analyzed using the Methylation Specific Multiplex Ligation Probe Amplification (MS-MLPA), with a panel of 24 different tumor suppressor genes related to cell cycle control, apoptosis regulation, DNA repair, cell adhesion and cell growth. Results: In the study cohort 47% of patients had cytogenetic alterations prior to AZA treatment, 4 with isolated -5/del(5q), 7 with isolated −7/del(7q), 4 with trisomy 8, 4 with not isolated -5/del(5q), 1 with trisomy 14, and 5 with complex cytogenetics. Methylation analysis showed that most patients (74%) had at least one methylated gene, but only 10% of patients displayed more than 3 methylated genes. The most frequently methylated genes were IGSF4 (28.3%), CDKN2B (24.5%), ESR1 (22.6%), CDH13 (17%) and CDKN1B (11.3%). The presence of a high number (≥2) of methylated genes (p=0.02), an adverse cytogenetics (p=0.03) or anemia (p=0.04) were independent prognostic factors associated with shorter overall survival. Moreover, the analysis of those patients displaying “no methylation”, patients with 1 methylated gene, patients with 2 methylated genes and those with more than 3 methylated genes, showed that as the number of methylated genes increases, the survival was shorter. The patients displaying the highest level of methylation (more than 3 genes), had a very short survival (median OS of 9.3 months, p

Description: Abstract 1776 Array-based sequence capture (Roche NimbleGen) followed by next-generation sequencing (Roche GS FLX Titanium sequencing platform) was used to analyze genetic variations in 93 genes relevant in CLL and two chromosomal regions: 13q14.3 and 17p13.1. CD19+ cells from 4 patients with CLL and 4 patients with other hematological malignancies (used as controls) were studied. A custom-made data analysis pipeline was used to annotate detected variants, including known single-nucleotide polymorphisms (SNPs), amino acid consequences, genomic location and miRNA binding sites. The enrichment assay followed by NGS allowed the detection of over 1600 variations/sample (median 1721, range 1618–1823). All putative variants were first compared with published single nucleotide polymorphism (SNP) data (dbSNP build 130) and most of the variants detected were identified as known SNPs. Overall, 10% of variants detected in each sample were variations not previously described. Interestingly, a 4bp insertion/deletion polymorphism (rs2307842) in the 3′UTR of HSP90B1, target site for miR-223, was detected. There is an increasing evidence suggesting that SNPs in the 3′UTR targeted by miRNAs (known as miRSNPs) are associated with diseases by affecting gene expression. We hypothesized that this ‘GACT’ deletion disrupts the binding site for miR-223 thereby increasing the translation of HSP90B1 and we confirmed that miR-223 regulates HSP90B1 expression by 3′UTR reporter assays. The relative luciferase activity of the construct with wild-type 3′UTR (WT-3′UTR) was significantly repressed by 31% following miR-223 transfection (p

Description: Chromosome 11q22.3 deletion (del(11q)) is one of the most common cytogenetic alterations in CLL and usually involves both ATM and BIRC3 genes. Concomitant mutations in ATM and/or BIRC3 in the remaining allele have been associated with poor survival. Despite the encouraging efficacy of novel agents targeting BCR and BCL2 pathways, del(11q) patients still have an inferior outcome and the development of resistance to these drugs has been increasingly reported. We therefore investigated the functional impact of del(11q) together with loss-of-function mutations in ATM and/or BIRC3 and whether CLLs harboring these alterations could benefit from novel combinatorial therapies. To address these questions, we used the CRISPR/Cas9 system to generate an isogenic CLL cell line to model del(11q) derived from HG3 cells by introducing two guide RNAs targeting the 11q22.1 and 11q23.3 regions. The presence of a monoallelic deletion (size ~17 Mb) was confirmed in 100% of the cells by FISH. Truncating mutations in ATM and/or BIRC3 were introduced in the remaining allele, generating HG3 del(11q) ATMKO, del(11q) BIRC3KO and del(11q) ATMKOBIRC3KO (three clones per condition). In addition, single ATMKO and BIRC3KO mutations, or the combination of both, were introduced into both HG3 and MEC1 CLL-derived cells (three clones per condition). Functional in vitro studies revealed that del(11q) BIRC3KO cells had increased growth rates compared to del(11q) BIRC3WT clones (P

Description: Background: In B-progenitor acute lymphoblastic leukemia (B-ALL) the identification of additional genetic alterations associated with treatment failure is still a challenge. Aims: 1.To identify genomic gains and losses in B-ALL at the time of diagnosis and to correlate these abnormalities with the genetic characteristics and the patients outcome. 2.To assess the prevalence and prognostic impact of genetic lesions in IKZF1, TP53, CRLF2, IL7R, PAX5, JAK2 and LEF1 genes in B-ALL. Methods: A total of 215 B-ALL patients were eligible for this study. 115 (53.5%) had less than

Description: Background: Chromosome 14q32 rearrangements involving the immunoglobulin heavy chain gene (IGH) affect less than 5% of chronic lymphocytic leukemia (CLL) patients. Their clinical course is aggressive and the outcome, worse than other CLL subtypes (Cavazzini et al, 2008; Gerrie et al, 2012). However, the biology of CLL showing IGH rearrangements (CLL-IGHR) is not completely defined. The identification of novel recurrent mutations in CLL by next generation-sequencing (NGS) has offered a more comprehensive view into the genomic landscape of the disease and improved the prognostication of CLL. Thus, mutational analysis might be especially useful in those patients with uncertain prognosis, such as those carrying IGH rearrangements. Aim: To analyze the mutational profile of CLL-IGHR patients by targeted NGS in order to improve our understanding of the genetic underpinnings of this subgroup. Methods: The study was based on 899 CLL patients, well characterized at cytogenetic, biological and clinical level, forty-two of them (4.7%) showing IGH rearrangements. Targeted NGS was performed in 231 CLL samples: 117 with 13q deletion, 27 with 11q deletion, 26 trisomy 12, 42 showing IGH rearrangements and the remaining 19 without any cytogenetic alteration. CD19+ B cells were isolated and DNA extracted. SureSelectQXT targeted enrichment technology and a custom-designed panel (MiSeq, Illumina), including 54 CLL-related and recurrent mutated genes, was carried out. The panel yielded 100x or greater coverage on 97% of the genomic regions of interest and the mean coverage obtained was 600x. Mutations were detected down to 3% allele frequency. Results: The mutational analysis of CLL-IGHR patients identified a total of 72 mutations in 32 genes. Seventy-one percent of patients (30/42) harbored at least one mutation. The most frequently mutated genes in this cohort were NOTCH1 (28.6%), POT1 (14.3%), TP53 (9.5%), SF3B1 (7%), BRAF (7%), EGR2 (7%), IGLL5 (7%) and MGA (7%), followed by BCL2, HIST1H1E and FBXW7 (4.8%), uncommonly mutated genes in CLL at these frequencies (Table 1). In fact, mutations in NOTCH1, BRAF, EGR2, BCL2, HIST1H1E and FBXW7 were significantly associated with CLL-IGHR patients (p=0.013, p=0.003, p=0.021, p=0.038, p=0.038 and p=0.021 respectively). In terms of time to the first therapy (TFT), CLL-IGHR had an intermediate-negative impact (median TFT=24 months) compared to the presence of cytogenetic alterations associated with good prognosis such as 13q deletions (median TFT〉120 months; p

Description: Introduction Hermansky-Pudlak syndrome (HPS) is an inherited platelet disorder characterized by bleeding diathesis, oculocutaneous albinism (OCA) and, sometimes, serious clinical complicationssuch as immunodeficiency, granulomatous colitis, and/or pulmonary fibrosis. Heterogeneous clinical symptoms and a large number of possible genetic culprits (10 HPS genes, 〉120 exons) complicate an unequivocal diagnosis of HPS. This study aimed to assess the clinical and platelet phenotype in ten patients with suspected HPS, and to identify the underlying genetic defects. Methods Ten patients from six families (F1 and F3 were Spanish, F2 was Turkish and F4, F5 and F6 were Portuguese) presenting with OCA (confirmed by skin biopsy) and bleeding diathesiswere included. Bleeding was evaluated by ISTH-BAT score. Phenotyping included, in patients with fresh blood samples available, platelet aggregation and ATP release, flow cytometry (FC), 14C-serotonin uptake and whole-mount electron microscopy (EM). Patients DNA was analyzed using two different targeted panels by high throughput sequencing (HTS). Sequence variants classification was performed according to ACMP recommendations. Results Patient characteristics are summarized in table 1. In F1, that had no history of consanguinity, there were two affected sisters. Patients 1 (P1) had several episodes of gastrointestinal bleeding (GI), which was attributed to granulomatous colitis. F2 is a consanguineous Turkish family, were P3 had severe rectal bleeding, requiring colectomy combined with ileostomy surgery. Pathological examination of the colon was reported as non-granulomatous colitis. Her older sister (P4) had exhibited dyspnea and shortness according to diffuse bilateral pulmonary fibrosis (BPF) diagnosis. In F3, P5 had been referred with acute GI bleeding secondary to angiodysplasia. In the non-consanguineous F4, HPS was first confirmed in P6, who showed blonde hair, nystagmus and low visual acuity; his older sister was diagnosed with HPS later, at the age of 56 years old (P7), because her OCA was masked using dark brown hair-coloring products. In P8, born from a non-consanguineous family (F5), HPS was suspected early in life, four months of age, upon recognition of OCA, nystagmus, deep visual deficiency and exotropia with compensatory torticollis. Lastly, in the consanguineous Portuguese family (F6), the two affected children (P9 and P10) had also showed a horizontal and torsional nystagmus and reduced visual activity. P10 also suffered from epilepsy and mild development delay. In phenotyping studies, the Spanish patients (P1, P2, P5) showed impaired platelet aggregation to mild agonists and reduced platelet dense granules by FC and EM. No platelet studies could be performed in F2. In Portuguese patients (F4, F5 and F6), the ATP release studies demonstrated a dense granule deficiency (Table 1). Molecular diagnosis was achieved, as a first-line approach, by means of HTS gene panels that revealed: a) F1 (P1 & P2) a homozygous deletion c.2054delC (p.P685L fs17*) in exon 13 of the HPS4, which had been previously reported in one Asian patient who showed BPF; b) F2 (P3 & P4): anovel missense homozygousvariant c.272T〉C (p.L91P) in exon 4 of the HPS4. Remarkably, the phenotype of the two Turkish sisters was different, with one having had severe GI bleeding requiring colectomy, and the other had developed BPF. C) F3 (P5): a novel heterozygous variant c.2464C〉T (p.R822*) in exon 13 of the HPS3 was detected; d) F4 (P6 & P7) and F5 (P8): here a nonsense variant c.307C〉T (p.Q103*) was identified in exon 5 of the DTNBP1, which was previously reported in a Portuguese patient. E) F6 (P9 & P10): these patients carried a novel five base pair duplication in the single exon of HPS6, c.60_64dup (p.L22R fs*33). Conclusions This study reports 10 new HPS patients, which demonstrates the heterogeneous nature of this syndrome and the complex phenotype-genotype correlations. The novel HTStechnology has facilitated the molecular diagnosis of HPS in these patients. Among the underlying molecular pathology, we identified a novel p.L91P variant in HPS4 that is associated with a severe clinical phenotype. Funding Gerencia Regional de Salud (GRS 1647/A/17), Fundación Séneca (19873/GERM/15), Instituto de Salud Carlos III (ISCIII, PI17/01966, PI17/01311,CB15/00055), Grupo de trabajo SETH and Instituto de Investigación Biomédica de Salamanca (IBSAL, IBY17/00006). Table Table. Disclosures No relevant conflicts of interest to declare.

Description: Clonal evolution is considered as a hallmark of progression in chronic lymphocytic leukemia (CLL). Next-generation sequencing technologies have expanded our knowledge of genetic abnormalities in CLL and enabled to describe marked clonal changes. The acquisition of driver mutations accompanied by selectively neutral passenger changes may be essential to understand the transformation from diagnosis to later more aggressive stages. However, the role of genetic mutations and clonal evolution during the clinical progression prior any therapy is still largely unknown. Longitudinal studies analyzing CLL patients repeatedly before intervening treatment are currently scarce. Patients and methods: We examined the exomes from 35 CLL patients in 2 time-points. Two groups of patients were characterized: (i)patients with progression (n=26) in which we analyzed samples taken from an early stable stage (inactive disease) and during clinical progression (active disease), but before treatment (median of time to first treatment=2.7 years); (ii)patients without progression with a stable inactive disease until last follow-up (n=9) (median follow-up=5.25 years). We also compared patients that gained new cytogenetic aberration detected by FISH in the 2nd time-point with those who did not. Sequencing libraries were prepared using TruSeq Exome Enrichment and sequenced by Illumina HiSeq1000 (84X). Somatic mutation calling was performed by a standardized bioinformatics pipeline. Thereafter, driver mutations were identified using the Cancer Genome Interpreter (https://www.cancergenomeinterpreter.org), a novel tool that identifies variants that are already validated as oncogenic and predicts the effect of the mutations of unknown significance. Results: We identified 397 somatic mutations in 364 different genes, ranging from 4 to 26 mutations per patient. Among them, 58 driver mutations were identified, being SF3B1 (6/35, 17.1%), TP53 (4/35, 11.4%) and NOTCH1 (4/35, 11.4%) the most common mutated genes. Comparing progressive vs. stable group, CLL patients with clinical progression showed a higher intra-tumoral heterogeneity than cases without progression (median of somatic mutations=14[4-26] vs. 9[5-14]). Comparing both tumoral time-points in the same patient, we identified a total of 11 acquired driver mutations and 7 mutations increasing its allele frequency in more than double in the 2nd time-point respect to the 1st one. All of them were detected in patients with clinical progression. Interestingly, TP53 and BIRC3 exhibited recurrently acquired mutations (detected each one in 2 cases). Three driver mutations in cancer genes not yet known for CLL (DHX9, GNAQ and HDAC2) were also acquired. Within CLL progressive patients (n=26), we observed clonal evolution characterized by acquired cytogenetic aberration in 9 cases. In patients with progression but no cytogenetic aberration gained at the 2nd moment (n=17), we detected that almost half of them (7/17) showed clonal evolution by acquired or doubled driver mutations. In the remaining patients with clinical progression but without any clonal evolution (n=10), 6 cases showed a driver mutation of CLL genes associated with bad prognosis (SF3B1, TP53, NOTCH1 or RPS15) already at first time-point. In the stable group (n=9), none acquired or doubled mutation was detected. However, clonal evolution characterized by acquired cytogenetic aberration was observed in 4/9 stable patients: two of them acquired 13q- whereas the other two acquired 11q-. Within stable patients without clonal evolution (n=5), we detected one case with a driver mutation in SF3B1 already at 1st time-point (follow-up=5 years). Conclusion: Clonal evolution represents a central feature of tumor progression in CLL. Our data show that the disease is evolving during time even in stable patients without any clinical signs of disease activity. In progressive patients, the disease evolution is accompanied by new appearance or accumulation of driver mutations and cytogenetic aberrations. Moreover, progressive patients that showed less or no changes during time bore typical CLL drivers at the first time-point. Funding: Seventh Framework Programme (NGS-PTL/2012-2015/no.306242); Ministry of Education, Youth and Sports (2013-2015, no.7E13008; CEITEC 2020 (LQ1601)); AZV-MZ-CR 15-31834A-4/2015 and TACR (TEO2000058/2014-2019); PI15/01471; Junta de Castilla y León (MHS). Disclosures No relevant conflicts of interest to declare.

Description: Introduction: Myelodysplastic syndromes (MDS) are hematological disorders at high risk of progression to acute myeloid leukemia (AML). Although, next-generation sequencing has increased our understanding of the pathogenesis of these disorders, the dynamics of these changes and clonal evolution during progression have just begun to be understood. This study aimed to identify the genetic abnormalities and study the clonal evolution during the progression from MDS to AML. Methods: A combination of whole exome (WES) and targeted-deep sequencing was performed on 40 serial samples (20 MDS/CMML patients evolving to AML) collected at two time-points: at diagnosis (disease presentation) and at AML transformation (disease evolution). Patients were divided in two different groups: those who received no disease modifying treatment before they transformed into AML (n=13), and those treated with lenalidomide (Lena, n=2) and azacytidine (AZA, n=5) and then progressed. Initially, WES was performed on the whole cohort at the MDS stage and at the leukemic phase (after AML progression). Driver mutations were identified, after variant calling by a standardized bioinformatics pipeline, by using the novel tool "Cancer Genome Interpreter" (https://www.cancergenomeinterpreter.org). Secondly, to validate WES results, 30 paired samples of the initial cohort were analyzed with a custom capture enrichment panel of 117 genes, previously related to myeloid neoplasms. Results: A total of 121 mutations in 70 different genes were identified at the AML stage, with mostly all of them (120 mutations) already present at the MDS stage. Only 5 mutations were only detected at the MDS phase and disappeared during progression (JAK2, KRAS, RUNX1, WT1, PARN). These results suggested that the majority of the molecular lesions occurring in MDS were already present at initial presentation of the disease, at clonal or subclonal levels, and were retained during AML evolution. To study the dynamics of these mutations during the evolution from MDS/CMML to AML, we compared the variant allele frequencies (VAFs) detected at the AML stage to that at the MDS stage in each patient. We identified different dynamics: mutations that were initially present but increased (clonal expansion; STAG2) or decreased (clonal reduction; TP53) during clinical course; mutations that were newly acquired (BCOR) or disappearing (JAK2, KRAS) over time; and mutations that remained stable (SRSF2, SF3B1) during the evolution of the disease. It should be noted that mutational burden of STAG2 were found frequently increased (3/4 patients), with clonal sizes increasing more than three times at the AML transformation (26〉80%, 12〉93%, 23〉86%). Similarly, in 4/8 patients with TET2 mutations, their VAFs were double increased (22〉42%, 15〉61%, 50〉96%, 17〉100%), in 2/8 were decreased (60〉37%, 51〉31%), while in the remaining 2 stayed stable (53〉48%, 47〉48%) at the AML stage. On the other hand, mutations in SRSF2 (n=3/4), IDH2 (n=2/3), ASXL1 (n=2/3), and SF3B1 (n=3/3) showed no changes during progression to AML. This could be explained somehow because, in leukemic phase, disappearing clones could be suppressed by the clonal expansion of other clones with other mutations. Furthermore we analyzed clonal dynamics in patients who received treatment with Lena or AZA and after that evolved to AML, and compared to non-treated patients. We observed that disappearing clones, initially present at diagnosis, were more frequent in the "evolved after AZA" group vs. non-treated (80% vs. 38%). By contrast, increasing mutations were similar between "evolved after AZA" and non-treated patients (60% vs. 61%). These mutations involved KRAS, DNMT1, SMC3, TP53 and TET2among others. Therefore AZA treatment could remove some mutated clones. However, eventual transformation to AML would occur through persistent clones that acquire a growth advantage and expand during the course of the disease. By contrast, lenalidomide did not reduce the mutational burden in the two patients studied. Conclusions: Our study showed that the progression to AML could be explained by different mutational processes, as well as by the occurrence of unique and complex changes in the clonal architecture of the disease during the evolution. Mutations in STAG2, a gene of the cohesin complex, could play an important role in the progression of the disease. [FP7/2007-2013] nº306242-NGS-PTL; BIO/SA52/14; FEHH 2015-16 (MA) Disclosures Del Cañizo: Celgene: Membership on an entity's Board of Directors or advisory committees, Research Funding; Jansen-Cilag: Membership on an entity's Board of Directors or advisory committees, Research Funding; Arry: Membership on an entity's Board of Directors or advisory committees, Research Funding; Novartis: Membership on an entity's Board of Directors or advisory committees, Research Funding.

Description: MDS with MF is not recognized as a distinct entity in 2008 WHO classification, and the prognostic role of MF remains controversial after several retrospective evaluations (Della Porta, 2009; Greenberg, 2012). In order to minimize inter-observer variability, a European Consensus (EC) has been reached for MF grade assignation (Thiele, 2005). IHC analysis of p53 protein BM accumulation has been proposed as prognosticator in MDS (Ramos, 2002), and more specifically in MDS with isolated 5q- (Jädersten, 2011; Saft, 2014) where it correlates with TP53 mutations. Several gene mutations have a prognostic role in MDS (Bejar,2011; Papaemmanuil, 2013; Haferlach, 2014), mutTP53 ranking as the most potent molecular prognosticator. By contrast, CALRmut, common in MPN, seems rare in MDS (Klampfl, 2013; Nangalia, 2013) albeit a focused evaluation taking into consideration EC grading has not been performed so far. The aim of this study was to evaluate the events that underlie MF changes in MDS by studying 77 MDS patients (52M/ 25F; median age 76, range 31-88; WHO-2008: RCUD 2, RARS 4, RCMD 40, RAEB-1 15, RAEB-2 7, 5q- 7, MDS-U 2), that were recruited prospectively since June 2006 and followed up to June 2104 in 7 GESMD centers. Diagnosis and cytogenetic evaluation followed GESMD SOP’s (Schanz’s cytogenetic score: Very good 6, Good 51, Intermediate 12, Poor 5, Very Poor 5). The study was approved by the IRB at each study site and all patients gave written informed consent. Median follow-up was 3.0 years (0.1-7.7). During this time, 49 patients (63.6%) died and 2 (2.6%) were lost to follow-up, while 20 (21.6%) progressed into acute leukemia. MF was evaluated independently by 2 pathologists following EC guidelines (MF-0 29, MF-1 31, MF-2 15, MF-3 2). IHC accumulation of p53 protein was evaluated as described previously and soluble p53 protein was analyzed in plasma by ELISA. Serum levels of 9 cytokines and chemokines were analyzed using cytometric bead arrays and a FACScanto flow cytometer (Becton-Dickinson). WT1 gene expression was evaluated by RT-PCR in PB and calculated as the WT1/GUSratio. The presence of mutations was evaluated in 67 patients. Sequencing was performed on a MISEQ NGS (Illumina) system using a panel of 111 genes previously related to MDS or MF. Sequencing results were analyzed by using the VariantStudio software package (Illumina), the threshold for mutation calling was set to 5%. Descriptive statistics and nonparametric statistical procedures were performed as appropriate, and univariate and multivariate analysis were performed for OS prediction. MF grade showed a statistically significant nonparametric correlation with the proportion of bone marrow erythroid precursors in BMA (-0.408), Hb levels (-0.271), number of PRBC transfused in the first 16 wks (0.302), ferritin levels (0.294), EPO levels (0.331) and p53 IHC score (0.231), as well as the presence of ALIPs (0.358), WT1/GUSratio (0.264) and serum levels of MIG (-0.251), but not with other covariates. Mutation incidence was highest for genes involved in splicing machinery (38.8%) as well as ASXL1 (18%). Mutated SF3B1 was associated with the absence of fibrosis (MF-0 23.1%, MF-1 7.4%, MF-2/3 0%; p=0.024) while the opposite was true for mutated ETV6 (MF-0 and MF-1 no mutations, MF-2/3 14.3%; p=0.041). Only 1 patient harbored a mutation in CALR. In our series, p53 protein accumulation and mutTP53 were not statistically linked. Mutations in GATA1 and SMC3 were more frequent in patients with IPSS-R High/Very High (25.0% vs. 3.6%, p=0.037 and 25.0% vs. 0%, p=0.005, respectively), but no single mutation was linked to Schanz’s cytogenetic score (Very Good/Good vs. Intermediate vs. Poor/Very Poor) with this sample size. Survival analysis showed MF-2 or higher as a relevant predictor for OS (median OS for MF-0/1 4.0 years [CI95% 1.4-6.6] vs. 1.2 [0-3.3] for MF-2/3; HR 2.3, CI95% 1.2-4.3, p=0.008), that was independent of IPSS-R in multivariate analysis (IPSS-R HR 2.47 [1.3-4.7], p=0.006; MF HR 2.18 [1.2-4.1], p=0.013). In conclusion, MF changes observed in MDS patients are not linked to mutations in CALR, cytogenetic score or IPSS-R, but to erythroid failure. Mutations in SF3B1 and ETV6 seem to diverge on their influence over MF. MF grade 2 or higher seems relevant for OS and independent of IPSS-R. Disclosures No relevant conflicts of interest to declare.