ALBERT

Description: Chronic lymphocytic leukemia (CLL) is characterized by the existence of subsets of patients with (quasi)identical, stereotyped B cell receptor immunoglobulins (BcR IG). Patients in certain major stereotyped subsets often display remarkably consistent clinicobiological profiles, suggesting that the study of BcR IG stereotypy in CLL has important implications for understanding disease pathophysiology and refining clinical decision-making. Nevertheless, several issues remain open, especially pertaining to the actual frequency of BcR IG stereotypy and major subsets, as well as the existence of higher-order connections between individual subsets. In order to address these issues, we investigated clonotypic IGHV-IGHD-IGHJ gene rearrangements in a series of 29,856 patients with CLL, by far the largest series worldwide. We report that the stereotyped fraction of CLL peaks at 41% of the entire cohort and that all 19 previously identified major subsets retained their relative size and ranking, while 10 new ones emerged; overall, major stereotyped subsets had a cumulative frequency of 13.5%. Higher-level relationships were evident between subsets, particularly for major stereotyped subsets with unmutated IGHV genes (U-CLL), for which close relations with other subsets, termed 'satellites', were identified. Satellite subsets accounted for 3% of the entire cohort. These results confirm our previous notion that major subsets can be robustly identified and are consistent in relative size, hence representing distinct disease variants amenable to compartmentalized research with the potential of overcoming the pronounced heterogeneity of CLL. Furthermore, the existence of satellite subsets reveals a novel aspect of repertoire restriction with implications for refined molecular classification of CLL.

Description: Background: Interpreting the pathogenic potential of an amino-acid changing single nucleotide variant (SNV) in a disease related gene can be challenging, especially for rare variants for which little or no information is available in clinical databases. In silico predictors, tools that predict the functional impact of an SNV algorithmically, can be useful in this scenario, and guidelines for variant interpretation recommend their inclusion in the interpretation process. Resources such as the dbNSFP database, which contains pre-calculated prediction scores for dozens of different algorithms, are readily available today. However, individual predictors rarely come to the same conclusion, and even for well-known disease causing SNVs results can be heterogeneous or even contradictory, which complicates their interpretation. Ensemble predictors such as REVEL, MetaLR/SVM or CADD combine the knowledge/information from multiple individual sources. These predictors use machine learning methods and training sets of pre-defined pathogenic and benign SNVs to integrate individual algorithms into a single, easy to interpret score. However, current training sets are based on pathogenic germline variants, which might cause these predictors to underperform when testing somatic variants. Aim: Development of HePPy (Hematological Predictor of Pathogenicity), an ensemble in silico predictor trained on somatic disease causing variants for use in a hematological setting. Methods: We followed the approach laid out by REVEL and used 10 in silico predictor scores and 4 phylogenetic conservation scores from the dbNSFP data base to train a random forest model. Our training set consisted of 371 unique missense SNVs from 61 hematologically relevant genes that were recurrently identified (in at least 10 patients) during routine diagnostics. All were consistently and unambiguously characterized by hematological experts as either a pathogenic somatic variant (n = 268) or a benign germline variant (n = 103) using a rigorous manual classification process within a data set of 69,879 cases studied between 2005 and 2018. Model accuracy was assessed by 10-fold cross-validation and further evaluated using a test data set consisting of 335 rare missense SNVs from routine diagnostics for which control germline material (buccal swabs, finger nail clippings) from the respective patients was available. Variants originating in the germline were expected to be mainly benign (n = 123), while somatic variants were considered pathogenic (n = 212). We compared the performance of this new tool to REVEL, MetaLR/SVM, CADD and the popular individual predictors SIFT and Polyphen2 by generating receiver operating characteristic (ROC) curves and calculating the area under the curve (AUC). Model implementation and analysis was performed using the R libraries "randomForest", "caret" and "pROC". Results: HePPy scores range from 0 (benign) to 1 (pathogenic) and cross-validation on the training set indicates a high accuracy of 0.968, which is also reflected by the clear separation in the distribution of obtained scores for benign and pathogenic training SNVs (see figure B). Application of the model to the test data set of rare SNVs shows that HePPy (AUC = 0.873) outperforms all other prediction tools in separating germline from somatic variants (see figure A). Surprisingly, both MetaLR (AUC = 0.717) and MetaSVM (AUC = 0.703) performed worse than the individual predictors SIFT (AUC = 0.794) and Polyphen2 (AUC = 0.821), while CADD (AUC = 0.831) and REVEL (AUC = 0.850) showed better performance. HePPy scores for somatic test variants were heavily skewed towards very high values (mean = 0.917). Germline variants had significantly lower scores (mean = 0.466), but their distribution was much more uniform than for somatic variants (see figure C). This suggests, to consider a significant proportion of the rare germline variants to have pathogenic potential. This is in line with the growing awareness of pathogenic germline variants and familial predisposition and emphasizes the importance of in silico predictions and other tools to replace the simple "tumor vs. normal" comparison. Summary: We developed HePPy, a new in silico ensemble predictor that is trained on 371 well-defined hematopathological somatic missense variants, which outperforms other currently available methods for in silico prediction in a hematological setting. Figure Disclosures Hutter: MLL Munich Leukemia Laboratory: Employment. Baer:MLL Munich Leukemia Laboratory: Employment. Walter:MLL Munich Leukemia Laboratory: Employment. Kern:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership.

Description: Background: Clonal chromosome aberrations in Philadelphia chromosome-negative metaphases (CCA/Ph-) occur in a subset of chronic myeloid leukemia (CML) patients. A shorter survival was reported for "non -Y" CCA/Ph- cases (Issa et al. Blood 2017). Besides -Y, the most frequent CCA/Ph- is +8, but a broad spectrum of other abnormalities can be found, including 7q-/-7, which is a typical aberration of myelodysplastic syndromes (MDS). In previous analyses we had shown an increased number of molecular genetic aberrations in CCA/Ph- compared to non-CCA/Ph- patients (Schnittger et al. ASH 2013, 2014). However, the clinical impact and the evolution of the mutation pattern is largely unknown. Here we extended follow-up and genetic characterization of the initial CCA/Ph- cohort. Aim: To determine the pattern of molecular mutations and their evolution To determine if mutations are part of the Ph+ or Ph- clone Patients and Methods: We included 52 CCA/Ph- patients (female: 25; male 27), with a median age of 58 [33-81] years, and a median BCR-ABL1/ABL1 ratio of 4.322% [0-58.088%] (N.A. for 4 pts.) at the time of initial CCA/Ph- detection. The following CCA/Ph- were present: trisomy 8 (n=26), other trisomies (n=4), -Y (n=7), del(7q)/-7 (n=4), others (n=7), two CCA/Ph- (n=4). We performed sequencing of myeloid gene panels on follow-up samples (1 to 3 per patient) on Illumina's MiSeq and NextSeq instruments (library preparation: 29-gene panel Thunderstorm RainDance [Lexington, MA] or 28-gene panel TruSeq [Illumina, San Diego, CA]). Data was analyzed with SeqNext (JSI Medical Systems, Kippenheim, Germany). Detected mutations were monitored on additional time points to determine variant allele frequency (VAF: mutated/all reads) development. A reference cohort of 47 patients with no sign of CCA/Ph- after MMR achievement was presented as part of our initial study (Schnittger et al. ASH 2014). Results: Cytogenetic monitoring was available over a median period of 31 [0-126] months for the CCA/Ph- and 26 [12-85] months for the reference cohort. Of the CCA/Ph- patients, 5/52 (10%) acquired additional typical aberrations as CCA/Ph- clone (incl. one -7), while in the reference cohort only one of 47 (2%) patients developed a -Y clone (n.s.). On the molecular level, we conducted a median follow-up of 72 [9-150] months for the CCA/Ph- cohort (mutations and BCR-ABL1/ABL1 ratio). Following the CCA/Ph- detection, somatic mutations were found in 30/52 (58%) patients (up to 4 per patient): ASXL1 (n=13), DNMT3A (n=10), TET2 (n=6), NRAS (n=3), RUNX1 (n=3), non-recurrent (n=8). The VAF of 7 mutations was strongly correlated to the BCR-ABL1 ratio and thus most likely present in the Ph+ clone. Mutations in ASXL1 were present in the Ph+ clone in five patients, of whom four never reached MMR, while 6/8 patients with ASXL1 mutations in Ph-independent clones achieved MMR under first- or second-line TKI therapy. Molecular genetic aberrations in Ph- cells were found in 23/52 (42%) CCA/Ph- patients, but only in 2 of 47 (4%) cases of the non-CCA/Ph- cohort (p

Description: The microenvironment provides essential growth and survival signals to chronic lymphocytic leukemia (CLL) cells and contributes to their resistance to cytotoxic agents. Pharmacologic inhibition of spleen tyrosine kinase (SYK), a key mediator of B-cell receptor (BCR) signaling, induces apoptosis in primary CLL cells and prevents stroma contact-mediated cell survival. This report demonstrates a role of SYK in molecularly defined pathways that mediate the CLL-microenvironmental crosstalk independent from the BCR. Chemokine and integrin stimulation induced SYK phosphorylation, SYK-dependent Akt phosphorylation, and F-actin formation in primary CLL cells. Inhibition of SYK by 2 pharmacologic inhibitors and siRNA-knockdown abrogated downstream SYK signaling and morphologic changes induced by these stimuli. CLL cell migration toward CXCL12, the major homing attractor, and CLL cell adhesion to VCAM-1, a major integrin ligand expressed on stromal cells, were markedly reduced by SYK inhibition. In combination with fludarabine, the SYK inhibitor R406 abrogated stroma-mediated drug resistance by preventing up-regulation of the antiapoptotic factor Mcl-1 in CLL cells. SYK blockade in CLL is a promising therapeutic principle not only for its inhibition of the BCR signaling pathway, but also by inhibiting protective stroma signals in a manner entirely independent of BCR signaling.

Description: Background: In 15-20% of CLL cases no aberrations are detected by chromosome banding analysis (CBA) and FISH due to limited resolution, lack of evaluable metaphases or presence of aberrations in loci not covered by standard-panel FISH probes. As reported in our previous study (Haferlach C. et al., ASH 2015, abs ID#79545), genomic arrays (GA) detected abnormalities in almost 20% of cases classified as normal by CBA and FISH and these showed an impact on time to first treatment (TTT) (Vetro C. et al., EHA 2016, abs ID# E1069). The CLL subgroup without abnormalities in CBA, FISH, and GA has not been characterised in detail, so far. Aims: 1) to describe CLL without abnormalities by CBA/FISH/GA by evaluating an extended gene panel, the IGHV mutation status and the B-cell receptor (BCR) stereotypy; 2) to determine prognostic impact of these factors. Patients and Methods: CLL diagnosis was based on cytomorphology and immunophenotyping according to standard guidelines. From a cohort of 1190 patients at diagnosis, 133 (11%) were selected based on normal karyotype by CBA, no abnormalities by interphase FISH with probes for 17p13 (TP53), 13q14 (D13S25, D13S319, DLEU), 11q22 (ATM), centromeric region of chromosome 12 and t(11;14)(q13;q32) (IGH-CCND1) and no abnormalities by GA (SurePrint G3 ISCA CGH+SNP Microarray, Agilent, Waldbronn, Germany). IGHV mutation status and BCR stereotypy were determined according to Agathangelidis et al., Blood 2012, and DNA sequencing was performed for the following genes: ATM; SF3B1; TP53; KLHL6; KRAS; MYD88; NOTCH1; NRAS; POT1; FBXW7; HIST1H1E; XPO1; ITPKB; MAPK1; BIRC3; BRAF; DDX3X; EGR2; RIPK1; RPS15; CND2. Results: Median age was 66 years (range: 33-83). Median follow-up was 5.6 years, 33 patients (25%) received treatment since genetic analyses. 10-year overall survival (OS) was 76% and median TTT was 9.2 years. Mutations were observed in 53 patients (40%): SF3B1 (n=17; 13%); NOTCH1 (n=10; 8%); KLHL6 (n=6; 5%); TP53 (n=6; 5%); ATM (n=5; 4%); XPO1 (n=4; 3%); FBXW7 (n=3; 2%); MYD88 (n=3; 2%); DDX3X (n=2; 2%); POT1 (n=2; 1.5%); ITPKB (n=1; 1%); KRAS (n=1; 1%); NRAS (n=1; 1%); and no mutation in RPS15, CCND2, MAPK1, EGR2, BRAF, HIST1H1E, RIPK1, BIRC3. 6 patients had 2 simultaneous gene mutations and 1 patient had 3 (i.e. NOTCH1, ATM and TP53). A mutated IGHV status (IGHV-M) was present in 100 patients (75%) and an unmutated IGHV status (IGHV-U) in 33 patients (25%). IGHV-U was related to both the occurrence of any gene mutation (p

Description: Background: Chronic myeloid leukemia (CML) cells can acquire resistance to tyrosine kinase inhibitors (TKI) that in ~40% of cases is due to acquisition of mutations in the ABL1 kinase domain of the BCR-ABL1 transcript. The p.T315I (c.944C〉T) mutation (mut) mediates resistance to most BCR-ABL1 TKIs (Imatinib, Dasatinib, Nilotinib and Bosutinib), whereas sensitivity to ponatinib has been demonstrated. Patients with p.T315Imut show a rapid increase in malignant cell burden and can progress to blast crisis. An earlier detection of the p.T315Imut may allow TKI treatment intervention ahead of disease progression. However, the sensitivity of conventional Sanger sequencing for detection of mutations is not less than 10-20%. Aim: To study the dynamics of evolution and progression of the p.T315Imut using ultra-deep sequencing (UDS) in comparison with Sanger sequencing. Patients and Methods: We selected 18 CML patients with high p.T315Imut levels originally detected by Sanger sequencing for routine diagnostics. Subsequently, we backtracked prior blood samples of all patients for a mean period of eight months (2-15 months) before detection of p.T315Imut by Sanger sequencing, analyzing 3-7 time points per patient. Patients (4 female and 14 male) had a median age of 60 years (18-84 years) and received treatment as follows: only Imatinib (n=3), only Nilotinib (n=3), only Dasatinib (n=1), treated with two prior (n=6) or three prior TKIs (n=5) by the time of p.T315Imut detection by Sanger sequencing. For more sensitive mutation detection, we amplified the BCR-ABL1 fusion transcript and designed two sequencing amplicons (550 bp and 575 bp) for UDS with the XL+ Kit for extended read length (Roche/454, Branford, CT). A minimal read coverage of 1,000 per base was reached. Our backtracking study by UDS was performed on samples sent in at intervals of approximately 3 months. Results: To prove high sensitivity of UDS with the 454 XL+ protocol we performed dilution experiments for three sequence variants and replicated sequencing experiments with low level mutations. The detection limit was at 1-2% mutation level and thus is 10-fold better than the sensitivity reached by Sanger sequencing. At the time point of initial routine diagnosis of p.T315Imut the median mutation load was 87.5% (30-100%) by Sanger sequencing and very similar by UDS (median: 84%; range: 40-99%; R2=0.7). In 6/18 patients backtracking identified a sample with a low p.T315I mutation level of 10% (Sanger sequencing detection level) was reached. Thus, in 33.3% of all cases a small, early clone of CML with p.T315Imut was identified. At subsequent time points, all 6 patients experienced a strong increase of the p.T315Imut level (〉50%), which represents the very fast expansion of the mutated clone. In a second subset of 10 patients, the p.T315Imut load was already 〉30% when first detected by UDS. The median interval to the last p.T315I negative time point was 2.4 months (0.9-3.5) and no sample between the p.T315I negativity and high positivity was available. This subset confirms the fast outgrowth of the p.T315Imut positive clone. The p.T315Imut load had a median increase of 0.9% (0.2-3.1%) per day, when calculated as average increase from the last negative sample to the time point with maximum mutation load. The other 2 patients had high p.T315Imut levels (〉40%) for our entire monitoring period. At the time of p.T315I detection by UDS, we observed eight patients with additional resistance mutations. The accumulation of mutations in one clone results in an extremely resistant CML. This was detected in one patient, where a p.T253H clone (Imatinib and Dasatinib resistant) gained the p.T315Imut. This clone expanded to 73% within 79 days. In contrast, we identified five cases with multiple CML clones carrying different mutations. However, the p.T315Imut clone was able to overgrow up to six other resistant clones. Conclusions: We showed: 1) the p.T315Imut rapidly increases upon occurrence, supporting the relevance of regular mutation monitoring in CML patients, when resistance to TKIs is suspected. 2) that small p.T315Imut clones in the 1-2% range can be sensitively detected by UDS in 33% of all samples if sampling intervals are within the 3 months range. 3) earlier detection of the p.T315Imut by UDS is a potentially valid method to allow a prompt change of TKIs before clonal expansion of the p.T315Imut cells. Disclosures Baer: MLL Munich Leukemia Laboratory: Employment; ARIAD Pharmaceuticals: Research Funding. Kern:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Mariathas:MLL Munich Leukemia Laboratory: Employment. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Schnittger:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership; ARIAD Pharmaceuticals: Research Funding.

Description: Background: In AML patients (pts), pre-leukemic mutations in DNMT3A, TET2 and ASXL1 (DTA) were shown to persist in remission, which was not associated with survival (Jongen-Lavrencic et al. NEJM 2018). On the other hand Abelson et al. (Nature 2018) recently identified a specific pre-leukemic mutational spectrum in pts that eventually develop AML. NPM1 mutation (NPM1mut) identifies a WHO AML entity which accounts for about 30% of all AML. The absence of NPM1 transcripts following treatment defines complete molecular remission (CMR). We aimed to dissect the clonal hierarchy of co-mutations at diagnosis of NPM1mut AML and analyze the role of persistent mutations in this well-defined CMR setting. Methods: We investigated 136 pts with diagnosis of NPM1mut AML (from 2006 to 2016) who attained a CMR (absence of NPM1 transcripts, sensitivity 0.001%) after intensive treatment. The mean follow-up was 40 months (2-96). Next-generation sequencing of 39 genes associated with myeloid malignancies was performed for all pts in paired samples at diagnosis and at CMR. The median coverage was 5200x. FLT3-ITD was analyzed by gene scan. Results: At diagnosis, a total of 375 mutations were detected, with an average of 2.8/pt (range 1-6); at least one mutation other than NPM1 was present in 123/136 pts (90%). DTA-genes DNMT3A (39% of pts) and TET2 (15%) were among the most frequently mutated, while no ASXL1 mutation was observed. Mutated non-DTA genes included: FLT3-ITD (35%), IDH2 (27%), IDH1 (21%), NRAS (18%), FLT3-TKD (14%) and PTPN11 (13%). The mean variant allele frequency (VAF) was 36% (2.7-94), and was higher for DTA than non-DTA mutations (44% vs 34%, p

Description: Background: Genome instability is a hallmark of cancer. Mutations in DNA repair pathway genes are frequent in a number of solid tumors. Defects in DNA repair or damage response can weaken response to conventional chemotherapy and are frequently regarded as poor prognostic markers. However, a high tumor mutation burden (TMB, number of somatic mutations per mega base) was recently found to correlate with better response to immune checkpoint inhibitors e.g. in colon cancer. Patients with defects in the DNA mismatch repair (MMR) pathway in solid tumors are among the cases with the highest TMB. Hematological malignancies are generally expected on the lower end of the TMB spectrum. We used whole genome sequencing (WGS) for 3256 patients with hematological malignancies (lymphatic and myeloid) to determine factors of genetic instability across all entities. Aim: To determine the number of known mutations in genes from the DNA repair pathway To estimate TMB using WGS and identify cases with high TMB in hematologic malignancies Methods: We investigated a cohort of 3256 patients with hematological malignancies, who were analyzed according to WHO diagnostic gold standards for routine purposes (incl. 584 acute myeloid leukemia [AML] and 635 myelodysplastic syndromes [MDS] samples). We performed amplification-free library preparation and sequencing on HiseqX and NovaSeq 6000 with a median coverage of 106x. Mapping and variant calling was performed with standard pipelines via BaseSpace (all Illumina, San Diego, CA). A pool of gender-matched genomic DNA (Promega, Madison, WI) was used for a tumor-unmatched normal variant calling. (a) In detail we evaluated 180 genes involved in DNA repair. We filtered on (likely) pathogenic variants from ClinVar and for TP53 on protein-truncating variants and (likely, possibly) pathogenic variants from the IARC database. (b) For TBM calculation we determined protein-altering changes and then subtracted all gnomAD listed variants in order to eliminate most germline variants. Results: We found 479 of 3256 (15%) patients with at least one pathogenic variant according to current database annotations in DNA repair or damage response genes. Most pathogenic variants were found in TP53 (330/3256; 10%) and ATM (25/3256, 1%), however, this is probably the effect of the already available systematic database annotation for both genes. For routine diagnostic purposes TP53 mutation status had been analyzed for 1184 patients with Sanger sequencing (7%) or amplicon next-generation sequencing (93%). A 98% and 99% concordance of the pathogenic and non-pathogenic TP53 status was found in comparison to WGS. Mutations in genes from the DNA double-strand break repair (and/or homologous DNA pairing and strand exchange) pathway were found in 93 patients (3%). Pathogenic and potentially germline MMR gene mutations were found in only 3 patients (0.1%, 2 MLH1, 1 MSH6), which equals the expected frequency in the Western population (0.05-0.3%). Next, we calculated TMB. The average was 2.4 [range: 0.4-39.2]. Only samples above the 95th percentile were defined as "TMBhi" (TMB ≥5). TMB was lowest in chronic myeloid leukemia (CML) and essential thrombocythemia (ET) (〈 2) and no ET or CML patient was found among the TMBhi. We then focused on MDS, which is our largest subcohort: 56 of 635 (9%) patients were in TMBhi. Furthermore, among MDS patients, a significantly higher TMB was found in MDS-EB-2 (average 3.3 vs. 2.3 for non EB-2, p

Description: Background: Diagnostics of hematologic malignancies is based on an integrated application of various methods. Different methods targeting the same structures, e.g. cytomorphology and flow cytometry both applied to identify and quantify pathologic cell populations, are aiming at revealing consistent results. While in the majority of cases achievement of such consistency is straight forward and in other cases both methods complement each other the alignment between cytomorphologically defined blasts and flow cytometrically defined myeloid progenitor cells (MPC) sometimes is challenging, in particular in myelodysplastic syndromes (MDS), acute myeloid leukemia (AML) and related diseases (thresholds defined for cytomorphology in WHO classification). We therefore identified such cases with higher percentages of cytomorphologically defined blasts than flow cytometrically defined MPCs and analyzed their genetic background in comparison to MDS cases with consistently low percentages by both methods and to acute myeloid leukemia (AML) cases with consistently high percentages by both methods. Aim: To clarify divergent findings between cytomorphology and flow cytometry on blasts and MPCs in patients with MDS and related diseases by analyzing the genetic background. Patients and methods: We identified 49 cases analyzed for myeloid malignancies, which were found to have higher percentages of cytomorphologically defined blasts than flow cytometrically defined MPCs (group 1). For comparison 83 patients with AML (group 2) and 53 cases with MDS (group 3) were selected in which percentages of cytomorphologically defined blasts (%B) were matching flow cytometrically defined percentages of MPCs (%MPC). All samples analyzed were bone marrow aspirates. Patients´ ages ranged from 40 to 60 (median 74), 23 to 89 (69) and 46 to 93 (76) years, respectively, in groups 1, 2 and 3 and sex distribution was 26/23 (f/m), 35/48 and 20/33, respectively. The median %B were 24% (range 11 to 62), 46% (20 to 98) and 5% (0 to 18), respectively, for groups 1, 2 and 3. The respective figures for %MPC were 9% (0.1 to 18), 52% (20 to 94) and 3% (0.3 to 16), respectively. The three groups were compared regarding cytogenetics based on chromosome banding analysis as well as regarding NGS data on mutations in 34 genes associated with myeloid malignancies. Results: Normal karyotypes were present in 37/49 (76%), 42/83 (51%) and 34/53 (64%) cases, respectively in groups 1, 2 and 3 (p=0.02). The respective figure for complex karyotypes was 5/49 (10%), 15/83 (18%) and 5/53 (9%, n.s.). Other chromosomal aberrations occurred less frequently and did not differ significantly between groups. Thus, karyotypes observed in group 1 were more similar to MDS than to AML. Regarding mutational profiling group 1 was characterized by a mutation spectrum similar to MDS developing into secondary AML while AML cases displayed a different spectrum as anticipated. Thus, in group 1, as compared to groups 2 and 3, the following gene were found more frequently mutated: ASXL1 (45% vs. 16% vs. 30%, p=0.001), SRSF2 (43% vs. 24% vs. 23%, p=0.042), TET2 (36% vs. 13% vs. 29%, p=0.014), RUNX1 (27% vs. 15% vs. 17%, n.s.). Interestingly, these genes were mutated at the lowest frequencies in group 2 most probably reflecting the low proportion of AML evolved as secondary disease after MDS. In contrast, SF3B1, which confers a favorable prognosis in MDS, was present at low frequencies in both groups 1 and 2 and at higher frequency in group 3 reflecting the low rate of AML progression in MDS with SF3B1 mutations (8% vs. 4% vs. 21%, p=0.005). Matching these findings mutations in genes typically associated with de novo AML were found most frequently in group 2 and only at low frequencies in groups 1 and 3: NPM1 (6% vs. 22% vs. 0%, p