U2AF1 Mutations Research Articles

Characterization of Clinical, Molecular, and Prognostic Features of the W.H.O. 2022 Classification System for Myelodysplastic Neoplasms (MDS)

Introduction: Myelodysplastic neoplasms (MDS) are clonal disorders of relative bone marrow failure and a variable risk of progression to acute myeloid leukemia. MDS generally exhibit genetic and/or cytogenetic abnormalities, an observation that led to the reclassification of MDS in the World Health Organization (WHO) 2022 update (Khoury et al, Leukemia 2022). In parallel, studies examining the mutational landscape of MDS have shown that specific patterns of somatic gene alterations strongly impact prognosis (Bernard et al, NEJM Evidence, 2022). In this study, we characterized the mutational patterns and clinical outcomes associated with the WHO 2022 classification schema to provide a more comprehensive view of this disease framework. Methods: This is a single-center retrospective cohort study of patients (pts) with clinicopathologically-confirmed MDS between January 2018 and December 2021. We identified cases with next-generation sequencing (NGS) performed on either bone marrow (BM) aspirate or peripheral blood (PB) using a clinically validated 164 gene panel designed to detect genes recurrently mutated in both myeloid and lymphoid neoplasms. For all cases, demographic information, clinical and molecular data, WHO 2022 diagnostic class, and survival outcomes were collected. IPSS-R and IPSS-M scores were calculated. Survival was estimated using the Kaplan-Meier method. Comparisons of mutational frequencies between groups were made via Fisher's exact test. This study was approved by Stanford University's Institutional Review Board. Results: Our inclusion criteria identified 118 MDS pts with clinicopathologic and molecular review. The median age of the cohort was 72 years, 67% were male, and 70% were Caucasian. For morphologically defined pts, MDS with low blasts (MDS-LB) was the most common subtype (36% of the entire cohort), followed by MDS with increased blasts-2 (MDS-IB2; 24%) and MDS with increased blasts-1 (MDS-IB1; 16%). For genetically defined pts, MDS with low blasts and SF3B1 mutation (MDS-SF3B1) was the most common subtype (19%) while MDS with biallelic TP53 mutations (MDS-biTP53) comprised 6% pts (Table 1). Within the latter group, 86% had a complex karyotype. A co-mutation plot to visualize the distribution of mutations among the various subgroups is shown in Figure 1. Notably,TP53 mutations (either mono- or bi-allelic) occurred in 16% of the cohort, were frequent in pts with complex karyotype (17/22; 77%), and were exclusive of ASXL1, IDH2, or SF3B1 mutations. In the morphologically defined subgroups, pts with MDS-LB had the highest prevalence of TET2 mutations (12/42 pts; 29%), while pts with MDS-IB1 had none (0/19 pts). U2AF1 mutations were significantly more common in pts with MDS-LB (13/42; 31%) relative to pts with MDS-IB2 (2/28; 7%) (p = 0.04). Conversely, patients with MDS-IB2 more frequently harbored TP53 mutations compared to pts with MDS-LB (p = 0.002). Calculation of IPSS-M scores stratified 47% of pts into the lower risk categories, with the remainder in higher risk categories (Table 1). The median IPSS-M score for the entire cohort was 0.105 [min: -2.11, max: 4.59] (Moderate High, MH). Median IPSS-M scores were highest in pts with MDS-biTP53 (3.18; [-0.480, 4.59]) and MDS-IB2 (1.65; [-0.880, 3.47]) (Very High, VH), and lowest in pts with MDS-SF3B1 (-1.13; [-2.06, 0.250]). IPSS-R scores generally paralleled IPSS-M scores, with 46% of all pts falling into the lower risk groups (Table 1). Pts with MDS-biTP53 had a median IPSS-R score of 8.0 [3.0, 9.5] (VH) while pts with MDS-SF3B1 had a median score of 2.5 [1.5, 4.0] (MH). Median overall survival (mOS) from the time of diagnosis was highest for those with MDS-LB (not reached), MDS-IB1 (not reached), and MDS-SF3B1 (7.0 years). Survival for patients with MDS-IB2 and MDS-biTP53 was poor, with mOS of 1.5 years and 0.8 years, respectively (Table 1). Conclusion: Our data demonstrated the clinical and molecular characteristics of MDS pts evaluated within the WHO 2022 schema and showed that differing mutational features were prominently associated within the morphologically as well as the genetically defined subgroups. Further, OS differed substantially among the defined subgroups. Our incorporation of the IPSS-M and IPSS-R provides additive prognostic and survival components into the WHO 2022 classification, which together enhance this framework's utility for studying MDS and its natural history. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Effect of Mutation Allele Frequency on the Risk Stratification of Myelodysplastic Syndrome Patients

Introduction Myelodysplastic syndrome (MDS) is a diverse group of clonal myeloid neoplasms. The accumulation of mutations in stem cell compartments is involved in MDS pathogenesis. Mounting evidences demonstrate that the addition of mutation data improves the prognostic stratification in MDS patients. In addition to mutational profiles, the variant allele frequency (VAF) of individual mutations also influence the prognosis in MDS patients. Materials and Methods A total of 698 primary MDS patients were recruited. The diagnoses were based on the 2016 World Health Organization classification. Patients with antecedent chemotherapy/radiotherapy or hematologic malignancies were excluded. TruSight myeloid sequencing panel (Illumina, San Diego, CS, USA) and the HiSeq platform (Illumina, San Diego, CA, USA) were adopted to analyze the gene alterations and mutant allele burdens. Because of the sequencing sensitivity issue, we verified CEBPA mutations and FLT3-ITD via Sanger sequencing. Results Exploration of the correlation between prognosis and VAF of individual genes showed that compared with wild-type genes, high VAF of mutations in 9 genes, including DNMT3A (cutoff value 40%, hazard ratio HR 1.67, p<0.001), TET2 (45%, HR 1.60, p<0.001), ASXL1 (20%, HR 1.48, p<0.001), E.ZH2 (40%, HR 1.45, p=0.037), SETBP1 (15%, HR 1.39, p=0.025), BCOR (80%, HR 1.57, p=0.045), SRSF2 (50%, HR 1.89, p=0.002), ZRSR2 (60%, HR 1.70, p<0.001) and TP53 (25%, HR 2.79, p<0.001) were significantly associated with shorter leukemia-free survival (LFS). With the exception of EZH2, SETBP1, and BCOR mutations, high VAF of all other six mutations were also associated with shorter overall survival (OS). In addition, the patients with SRSF2 and TP53 mutations, even at lower VAF, had poorer prognosis for both LFS and OS compared with those with wild-type genes. For other 7 above-mentioned mutations, only higher VAF conferred poorer outcomes; the patients with lower mutation VAF had outcomes similar to the patients with wild-type genes. In multivariable analysis (Table 1), female sex and receiving allogenic hematopoietic stem cell transplantation (HSCT) were independent favorable factors for both LFS and OS, while older age and higher-risk IPSS-R predicted shorter LFS and OS. Mutant IDH2 and CBL predicted both shorter LFS and OS, while U2AF1 mutation and DNMT3A and ZRSR2 mutations with high VAF were associated poorer OS. The mutation status and VAF levels of TP53 mutations could distinguish patients into three hierarchical groups with different prognoses. The presence of poor-risk mutations as identified in the multivariable Cox analysis could re-stratify the IPSS-R risk groups. Patients with these unfavorable mutations in each IPSS-R risk subgroup had survivals worse than other patients of the same risk but similar to or even worse than those in the next higher-risk subgroup (Figure 1). Considering OS, the incorporation of the molecular data in the IPSS-R could reclassify 8.9% (18/202) of IPSS-R very low/low-risk patients to intermediate-risk subgroup, 17.9% (33/184) of IPSS-R intermediate to high-risk subgroup, and 34.4% (54/157) of IPSS-R high to very high-risk subgroup. Patients harboring U2AF1 mutation had poorer LFS and OS if they were not treated with hypomethylating agents (HMA), but had similar LFS (wild-type vs. mutant-type, 14.3 vs. 15.5 months, p=0.537) and OS (wild-type vs. mutant-type, 22.2 vs. 40.1 months, p=0.450) compared with those with wild-type U2AF1 if they received HMA treatment. Moreover, patients with low TP53 mutant VAF got benefit from HMA treatment considering LFS. Conclusion We provide evidences that VAF is critical for risk stratification in MDS patients and should be considered in novel scoring systems. Our data fostered our understanding of the mutation burden of the diseases and provided future patient-tailored therapeutic avenues. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Impact of Spliceosome Mutations on the Prognosis of Myelodysplastic Syndrome (MDS)

Introduction: More than half of patients with myelodysplastic syndrome (MDS) have spliceosome mutations (SM) including ZRSR2, U2AF1, SF3B1 and SRSF2. Splicing is fundamental for proper mRNA processing. Mechanistic descriptions of the after-effects of SM remain elusive as improperly formed mRNA quickly decays or changes are subtle and may lack observable effects. These mutations have been implicated in multiple biological alterations including myeloid proliferation through aberrant NF-kB signaling. SRSF2P95* has been consistently shown to provide the highest fitness advantage in the bone marrow environment. This study reports the characteristics, treatment outcomes, leukemia free and overall survival of MDS with spliceosome mutations. Patients and Methods: A retrospective review of patients diagnosed with MDS older than 18 for whom genomic data was available were included at a single institution. Patient characteristics and bone marrow data, including cytogenetic and next generation sequencing (NGS) information were assessed. Genomic DNA was extracted from whole bone marrow aspirate samples and subject to 81-gene target PCR-based sequencing using a NGS platform. The IWG 2006 response assessment criteria were used. Overall and leukemia-free survival were calculated, the latter as a composite of acute myeloid leukemia diagnosis or death from any cause, using the log-rank test for group comparison. Results: First, we compared patients with SM vs no SM. We identified 427 patients with SM and 658 patients with no SM from 2017 to 2022 (Table 1). Among patients with SM, 71% were male, median age was 71. SM patients had higher incidence of IPSS Low (33.3% vs 10.3%) and IPSS Int- 1 (35.6% vs 32.2%), and a higher incidence of diploid karyotype (45% vs 21%), compared to patients with no SM. There was no difference in treatment regimens, and 66.9% of patients with SM received hypomethylating agents (HMA) alone or in combination vs 60.3% in the group without SM. MDS with SM had a lower CR/CRi rate (27% vs 36%); however, they had higher hematologic improvement (HI) (23% vs 14%) and lower chance of disease progression or death at the time of response assessment (39% vs 31% (p = 0.01)). Leukemia-free survival was longer in patients with SM vs no SM (median 4.1 years vs 1.9 years (p<0.0001), this was independent of SF3B1 mutations. Median OS was 4.5 years vs 1.9 years in patients with SM and no SM, respectively (p <0.0001). We then compared outcomes between patients with specific SM mutations. These included SF3B1 (34%), SRSF2 (26%), U2AF1 (25%), and ZRSR2 (14%). Thirty-six percent were female. Median age was 71 (range 28-93). Median percentage of blasts was 3 (range 0-19). Most frequent diagnoses were MDS-EB (35%), MDS-RS (25%) and MDS-MLD (23%). IPSS was Low, Intermediate 1, Intermediate 2 and High in 33.3%, 35.6%, 17.1% and 4.2%, respectively. Fifty-seven percent were in the favorable cytogenetic risk group, 44% with diploid karyotype. Patients with a SM mutation did not share any other SM mutation (Table 2). There were no differences in therapy selected or response rate among SM mutations. Median LFS was not reached (NR), 5.7 years, 3.4 years, and 2.7 years for patients with SF3B1, ZRSR2, U2AF1, and SRSF2 mutations, respectively. Median OS was NR, 5.7 years, 3.6 years, and 2.7 years (p = 0.005) for the same groups, respectively. Conclusions: In our study, patients with SM had a lower rate of CR/CRi with HMA-based therapy, however, leukemia free and overall survival were markedly different for each mutated gene. SM mutations were significantly more prevalent in males. Compared to MDS with diploid cytogenetics, ZRSR2 and SF3B1 mutations have a similar median LFS and OS, but SRSF2 and U2AF1 mutations carry a significantly worse median LFS and OS. While it is convenient to group these mutations together, prognostication should be individualized to specific mutations and co-mutations. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Therapeutic Targeting of Spliceosome Mutant Myeloid Neoplasms Via PARP1 Inhibition

Somatic mutations in splicing factors (SF) SF3B1, SRSF2, U2AF1 are observed across a spectrum of clonal myeloid disorders including myelodysplastic syndromes (MDS). Previous studies identified increased R-loops in SF-mutant cells. Abnormal R-loop accumulation is associated with increased genomic instability and preferential sensitivity to ATR kinase inhibitor. Motivated by these observations, we set out to identify novel therapeutic targets in SF-mutant myeloid malignancies. Using MLLAF9 Srsf2WT and Srsf2P95Hcells, we performed a focused drug screen using inhibitors targeting different DNA damage response pathways and drugs that inhibit DNA replication or DNA metabolic processes. Unexpectedly, Srsf2P95H cells displayed increased sensitivity to inhibitors targeting PARP1/2 (Fig. 1A-C). Olaparib (PARPi) treatment induced higher apoptosis in Srsf2P95Hcells than in Srsf2WTcells (Fig. 1D). To ensure this phenomenon extends beyond Srsf2 mutation, we generated MLLAF9 leukemia on U2af1WT and U2af1S34Fbackgrounds. PARPi treatment in U2af1S34F cells resulted in reduced cell viability and increased apoptosis compared to U2af1WT cells (Fig. 1E-G). Next, we genetically inactivated Parp1 in Srsf2WT and Srsf2P95H cells. The absence of Parp1 did not affect the growth of Srsf2WT cells (Fig. 1H). In contrast, Parp1 deletion significantly reduced proliferation of Srsf2P95H cells (Fig. 1I).Finally, to confirm this phenotype in vivo, we transplanted established MLLAF9 Srsf2WTor Srsf2P95H leukemia into syngeneic mice. Eleven days post-transplant, mice received either vehicle or PARPi. Mice transplanted with Srsf2WT cells showed no survival advantage when treated with PARPi (Fig. 1J). In contrast, PARPi treatment significantly prolonged survival of mice transplanted with Srsf2P95H cells compared to vehicle-treated mice (Fig. 1K). To examine whether PARPi sensitivity is conserved in humans, we treated K562 cells expressing either SRSF2P95H or SF3B1K700E mutation. SRSF2P95H and SF3B1K700E cells showed increased sensitivity to olaparib and rucaparib (Fig. 1L-M). This observation was further validated in a panel of primary AML blasts carrying SRSF2 and U2AF1 mutations(Fig. 1N). Taken together, our results identify PARP1 as a novel therapeutic target in SF-mutant myeloid neoplasms. PARP inhibitors preferentially kill cells with defective homologous recombination (HR)-mediated DNA repair. We asked if the observed PARPi sensitivity in SF-mutant cells was due to defects in HR. To measure HR activity, we performed a CRISPR-Cas9/mClover assay. The assay measures HR-dependent insertion of an mClover-containing donor cassette into Cas9-mediated double-stranded breaks in the LMNA gene, which results in the expression of a green fluorescent Lamin A/C protein. There was no significant difference in the relative percentage of mClover positivity in SRSF2P95H or U2AF1S34F cells compared to control (Fig. 1O), suggesting that these cells are HR-proficient. Interestingly, SF3B1K700E cells showed significantly reduced HR efficiency, consistent with a published study. We next investigated cellular PARP1 activity in SRSF2- and U2AF1-mutant cells by monitoring Poly-ADP-ribosylated chain (PAR) levels. SRSF2P95H cells had elevated PAR level compared to SRSF2WT cells (Fig. 1P). Since PARP1 is known to associate with R-loops, we wanted to test whether increased PARP1 activity and PARPi sensitivity are R-loop dependent. We generated SRSF2-mutant cells that inducibly express RNaseH1, an enzyme that specifically cleaves the RNA moiety within the RNA:DNA hybrids. Induction of RNaseH1 reduced PAR level and PARPi sensitivity in SRSF2P95H cells, suggesting that SRSF2-mutant cells elicit a PARP1 response to promote survival in an R-loop-dependent manner (Fig. 1Q). Lastly, since SF-mutant cells are also sensitive to ATRi, we tested if combining PARPi and ATRi can further sensitize SF-mutant cells. Combined PARPi + ATRi treatment induced synergistic killing of Srsf2- and U2af1-mutant cells compared to wildtype cells in cell viability and colony forming assays (Fig. 1R-S). In summary, this study provides a pre-clinical rationale for therapeutic targeting of PARP1 in SF-mutant leukemia. Moreover, PARP and ATR inhibitor combination could emerge as a new therapeutic strategy in this genetically distinct disease subtype. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Analysis of Clinical Features, Prognostic Characteristics and Gene Mutations of Primary Myelodysplastic Syndrome Patients with Myelofibrosis

Objective ：This study aims to analyze the clinical features, prognostic characteristics and gene mutations of the primary myelodysplastic syndrome with myelofibrosis (MDS-MF) patients, ultimately to improve the cognition of MDS-MF. Methods ：From January 1, 2013 to February 1, 2020, 417 newly diagnosed primary MDS patients in the first affiliated hospital of Zhejiang University with bone marrow (BM) biopsy examination were included. They were divided into two groups according to whether BM associated with fibrosis (2005 European Myelofibrosis Network criteria), the MDS-MF group(MF grades 1-3）and the MDS without MF group (MDS-MF0).The clinical features, prognosis characteristics and gene alterations were analyzed between the two groups retrospectively. Results ： ① MF was confirmed in 46.3％（193/417）cases of all the MDS patients, of which 66.3％（128/193）were MF-1 , 25.9％（50/193）were MF-2 and 7.8%(15/193) were MF-3 respectively. The proportion of MDS-EB in MDS-MF group was 48.2%(93/193) and significantly higher than in MDS-MF0 group 35.7%（80/224）,the difference had statistical significance(P=0.008). However, the occurrence of MF does not differ among the MDS subtypes according to the 2016 WHO classification (P=0.342). ② The clinical features of the two groups had statistically significant differences in the higher number of BM blast cells(P=0.006), higher prevalence of death (P=0.031), higher rate of Leukemic transformation (P<0.001), higher level of serum lactate dehydrogenase (LDH) in the peripheral blood (p<0.001), higher level of β2-microgram in the peripheral blood(p=0.003) and higher level of C-reactive Protein (CRP) in the peripheral blood (P=0.004) in the MDS-MF group than in the MDS-MF0 group. The number of WBC, Hb, PLT and Ferritin at the first diagnosed in our hospital were no difference compared the two groups (All P>0.05). The sex, Age and the IPSS-R score(IPSS-R >3.5 v ≤3.5) were no difference in the two groups (ALL P>0.05). ③Compared with MDS-MF0 group, MDS-MF were significantly associated with worse OS(P=0.035) and worse PFS(p<0.001). The median overall survival was 13.20(0.07-90.70) months in MDS-MF group, and was 23.77(0.53-98.97) months in MDS-MF0 group respectively. Subgroup analysis showed that both OS and PFS of either the MDS-MF1-2 group or the MDS-MF 3 group were also significantly worse than those of the MDS-MF0 group（P＝0.013 and <0.001 respectively）. ④Among all the patients ,337 were finished Next-generation-sequencing (NGS) test, sequencing errors and all known or possible single-nucleotide polymorphisms were excluded. The number of mutations (>2) was more frequently in the MDS-MF, the difference had statical significance(P=0.030). The RNA splicing was more frequently in the MDS-MF than the MDS without MF, the difference also had statical difference(P=0.025). The ratio of U2AF1 mutation occurred in MDS-MF group higher than MDS-MF0 group, the difference had statical significance (P=0.027). There was totally 14.2% (48/337) patients with the mutation of U2AF1, of which 19.7%（29/147) in MDS-MF group and 10.0%（19/190) in MDS-MF0 group. Conclusion: MDS-MF has unique laboratory and clinical characteristics. MF is an independent risk factor for shorter overall survival (OS) and worse progression-free survival (PFS) in primary MDS. Evaluation of MF is very significant for prognosis judgment in primary MDS. The U2AF1 mutation may be a biomarker related to the occurrence and poor prognosis of MDS-MF. However, the mechanism of U2AF1 in MDS-MF patient needs to be further explored. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Favorable Outcomes of DDX41-Mutated MDS and Low Blast Count AML Treated with Azacitidine with or without Lenalidomide (ALLG MDS4 Trial)

Introduction. The Australasian Leukaemia and Lymphoma Group (ALLG) MDS4 phase II trial randomized 160 patients (pts) with higher risk (by IPSS) myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML) and low blast acute myeloid leukemia (AML) to either azacitidine (AZA) or combination AZA with lenalidomide (LEN, from cycle 3 onwards) (Kenealy, Haematologica 2019; ANZCTR12610000271000). Whilst the combination of LEN and AZA was tolerable, there was no improvement in clinical benefit, response rates or overall survival in patients compared to treatment with AZA alone. We performed a post-hoc genomic analysis of pts on this trial to further understand the molecular changes in this cohort and any correlation with outcomes. Methods. A unique molecular index (UMI) based QIAseq targeted DNA panel (QIAGEN) was performed as previously described (Blombery, Blood 2020). Sensitivity for variant calling was 0.5% variant allele frequency (VAF). HumanCytoSNP-12 BeadChip array (SNP-A) was previously performed and reported (Kenealy, Haematologica 2019). Primary endpoint of the study was ongoing clinical benefit at 12 months, defined as being alive and progression-free. Logistic regression and Cox proportional hazards regression were used respectively to assess the differences in response rates and overall survival between two treatment groups according to the molecular subgroups. Results. A total of 66 pts, representative of the 160 enrolled, had baseline gDNA available for targeted panel testing; 36 and 30 pts on AZA alone and AZA+LEN therapy, respectively. Median age was 71.6 years. IPSS-R risk groups were very low or low (n=20), intermediate (n=18), high or very high (n=25), and not available (n=3 due to missing karyotype). Disease subtypes were MDS (n=52), CMML (n=9), and AML (n=5). 47% of pts had an abnormal SNP-A, including 18% with complex (≥3) copy number changes. The most common gene mutations observed were ASXL1 (48.5%), TET2 (47%) and RUNX1 (30%) (Figure 1). Overall clinical benefit at 12 months was achieved in 39 pts (59%). TET2 was the only gene mutation associated with significantly improved benefit when treated with AZA alone (OR 10 [95% CI, 2 to 81]). Median follow up among survivors was 49 months and the median overall survival was 33.2 months (95% CI, 21.7 - NE). Mutations in TP53, U2AF1, EZH2 and SRSF2 were associated with inferior overall survival. Molecular subgroup analyses did not identify any significant difference between the two treatment arms. Given the recent discovery of DDX41 mutations in MDS/AML and early data supporting a favorable response to AZA/LEN, we focused on 6/66 (9%) patients with detectable DDX41 mutations. All 6 pts had a second mutation detectable in DDX41 as well as no detectable abnormalities on SNP-A. The most commonly co-mutated gene in this group was ASXL1 (in 4/6). One patient with DDX41 mutation died before receiving any therapy. Best responses achieved in DDX41 mutated patients were complete remission (CR, n=1), marrow CR (n=1), hematologic improvement (HI, n=1), and stable disease (SD, n=2). Remarkably, of the 5 patients that received treatment (4 AZA, 1 AZA+LEN), the overall survival at 52 months (range 47-58) follow up was 100% (Figure 2), despite having intermediate (n=3) and high (n=2) R-IPSS. Mutations were also serially assessed on therapy in 45 pts: ≤2 cycles in 38 pts, and ≥4 cycles in 7 pts (2 AZA+LEN). Of 166 mutations observed at baseline, majority (67%) were stable with <10% change in VAF (median +2%). Overall, 27 (60%) pts had clearance or >10% VAF reduction in ≥1 variant, including clearance in 20 (12%) variants (most commonly TET2 [n=4] and RUNX1 [n=3]), and reduction in 33 (20%) variants (ASXL1 [n=6], SF3B1, SRSF2 and TET2 [n=5 each], but no pt had complete molecular clearance. In contrast, new mutations (n=18 variants) or >10% VAF increase (n=2 variants) were observed in 14 (31%) pts, most commonly TET2 (n=6), RUNX1 (n=4), and DNMT3A and BRAF (n=3 each). Conclusions. In summary, our correlative molecular analysis of the ALLG MDS4 trial of AZA vs AZA+LEN in MDS/AML has identified, in addition to established molecular risk factors for inferior outcomes, that pts with DDX41 mutations are associated with highly favorable outcomes with either AZA or AZA+LEN. This observation and subgroup of patients warrants further study in randomized trials. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

MDS Classification - Do We Still Have to Count Blasts?

Background: Since the definition of MDS their classification relied on cytomorphology determining dysplasia, cytopenia and blast count. The revised 4th edition of the WHO classification encompasses only one genetically defined entity (MDS with isolated del(5q)). In the 5th edition of the WHO classification 2 additional entities are defined by genetics: MDS with low blasts and SF3B1 mutation and MDS with biallelic TP53 inactivation. All other are classified still based on morphology into i) hypoplastic MDS, ii) MDS with low blasts and iii) MDS with increased blasts. Aim: As blast counting is subjective and the boundary between low and elevated blasts is arbitrary, we set out to propose a classification for MDS based solely on genetic abnormalities obviating the need for blast counting. Patients and Methods: The cohort comprised 735 de novo MDS. The diagnosis was established following WHO classifications. In all cases chromosome banding analysis and WGS (100x, 2x151bp, NovaSeq) were performed. Variants were called with Strelka2, Manta and GATK using a tumor w/o normal pipeline. For all cases overall survival data was available (median follow-up: 9.3 years). Results: We categorized our cohort 1) according to WHO revised 4th edition, 2) 5th edition, 3) according to blast count (<5%, 5 to <10%, 10 to <20%), and 4) according to numbers of mutations. For all 4 categorizations prognostic differences between subgroups was confirmed. Next, we performed cluster analyses to identify genetic abnormalities associated with blast count. While SF3B1 mutations and isolated del(5q) were associated with low blast count, mutations in ASXL1 and RUNX1 were associated with higher blast count (for all p<0.0001). Further, we evaluated the genomic landscape for associations of genetic parameters. RUNX1 mutations and ASXL1 mutations were highly associated with spliceosome mutations. 49/62 (79%) RUNX1mut MDS and 97/153 (63%) ASXL1mut MDS harbored also spliceosome mutations. On the other hand RUNX1mut, ASXL1mut and spliceosome mutations rarely co-occurred with biallelic TP53 inactivation or complex karyotype (only 5/41 cases and 19/78 cases). 102/735 (14%) cases harbored neither a del(5q), a complex karyotype, biallelic TP53 inactivation nor a mutation in SF3B1, SRSF2, U2AF1, ZRSR2, DNMT3A, TET2, RUNX1 or ASXL1. In 20 of these at least one mutation was found in 21 additional genes known to be mutated in MDS. Based on these data 9 genetically defined non-overlapping hierarchical subgroups (definitions of the respective prior subgroups are exclusion criteria for the following ones) are proposed: 1. biallelic TP53 inactivation 2. complex karyotype, 3. RUNX1 mutated, 4. ASXL1 mutated, 5. del(5q), 6. SF3B1 mutated, 7. mutated DNMT3A, TET2, U2AF1, SRSF2, and/or ZRSR2, 8. presence of at least one mutation in one of 21 other genes recurrently mutated in MDS, 9. none of the genetic markers present that define any of the previous entities. Survival analyses revealed marked differences between these biologically defined entities (figure 1). In univariate cox regression analyses including all 9 genetic categories and blast count the categories biallelic TP53 inactivation, complex karyotype, RUNX1mut, ASXL1mut, SF3B1mut, none of the genetic markers and blast count <5% vs ≥5% and <10% vs ≥10% were significantly associated with OS (for all p<0.002). In multivariate analyses both blast count cut offs did not show a significant impact, while the categories biallelic TP53 inactivation, complex karyotype, RUNX1mut, ASXL1mut, SF3B1mut were independently associated with OS (for all p<0.05). In figure 2 the relationship of classification according to the 4th revised, the 5th edition of the WHO and the purely genetically based classification as proposed here is depicted. Conclusions: Based on karyotype and mutation status of 9 genes complemented by 21 genes in 10-15% of cases, MDS can be separated into 9 biologically distinct subgroups, reflecting biology better than blast counts. These are associated with significant differences in overall survival. RUNX1 and ASXL1 define distinct subgroups within spliceosome mutated MDS potentially driving progression. Given the poor survival of the MDS subgroups with biallelic TP53 inactivation, complex karyotype and RUNX1 mutation it has to be further evaluated whether or not these MDS subtypes should be combined with AML harboring the respective genetic abnormalities for treatment approaches. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

The Growth Kinetics of Clonal Hematopoiesis Inform Risk of Blood Cancer Progression

Introduction: Clonal hematopoiesis of indeterminate potential (CHIP) is a frequent consequence of aging associated with increased risk of blood cancers, although overall individual risk is < 1% per annum. Proposed risk factors for blood cancer progression include a variant allele frequency (VAF) > 10%, specific gene mutations or the presence of more than one CHIP clone. However, the relative value of monitoring CHIP clone growth with respect to blood cancer risk is unknown. To address this evidence gap, we measured CHIP at baseline and 3 years later in 11,298 healthy individuals aged ≥70y from the ASPREE randomized clinical trial (www.aspree.org). Methods: A custom, high sensitivity (0.5%) and reproducible amplicon-based targeted sequencing assay was designed to assess CHIP in 19 genes, except for SRSF2, which had poor coverage. CHIP was defined as a VAF ≥ 2% with a minimum of 5 variant reads for previously reported CHIP or 20 for variants not previously reported. The analysis involved measuring CHIP variants in two serial blood samples collected at baseline and year 3 of follow-up for longitudinal analysis. Associations between CHIP and blood cancers were assessed using Cox proportional hazards models. Results: CHIP at baseline was present in 34% of participants who had a corresponding year 3 sample with adequate sequencing depth. Unexpectedly, 1,245/3,110 (40%) CHIP clones present at baseline fell below 2%, while 1,520 new CHIP clones appeared at 3 years. This equated to an incident CHIP rate of 7% per annum. CHIP genes more likely to expand and contract were TP53 and ASXL1 (odds ratio 1.25 and 1.22 respectively). To measure growth characteristics of CHIP clones, we calculated a Fitness Score, FS (Log2(VAF3-year/VAFBaseline) for the 3,110 clones present at baseline. We observed no significant mean fitness advantage for any recurrent CHIP variant other than PPM1D R552* (Figure 1). CHIP variants of DNMT3A (including DNMT3A R882), TET2 and ASXL1 had a statistically significant negative FS. Even variants generally considered most pathogenic (SF3B1 K700E, JAK2 V617F, U2AF1 Q157, TP53 R273) had no significant fitness advantage. Some CHIP clones had a significant growth disadvantage (FS < -1). For example, only 1/28 DNMT3A S393N clones present at baseline remained above 2% and almost half of TP53 variants reduced in size by more than 2-fold at 3 years. We observed 299 blood cancers over a median time of 6.9 years. Individuals with CHIP clones >10% VAF (n= 652) had increased incidence of blood cancers (adjusted hazard ratio, aHR 2.14, confidence interval, CI 1.34-3.42, p=0.002). Genes associated with increased risk for blood cancers were SF3B1/U2AF1 (aHR 3.02, CI 1.63-5.61, p<0.001) and TET2 (n=aHR 1.89, CI 1.18-3.02, p=0.008). Participants with small clones (VAF 2-10%) at baseline or year 3 had no increased risk for blood cancers compared to those without CHIP. To investigate if the growth kinetics of CHIP predicts blood cancer risk, we subdivided CHIP clones into expanding (FS > +1, n= 104), stable (FS -1 to +1, n= 2043) and contracting (FS < -1, n= 963). Baseline VAF of expanding clones were smaller than stable clones: geometric mean 3.9% and 5.0%, respectively (p<0.001). Expanding clones were enriched 2-fold for PPM1D, SF3B1 and JAK2 V617F. Participants with expanding clones (n=99) were more likely to have more than one CHIP variant (50% compared with 39% for stable and 37% for contracting), although 90% of these additional variants were stable or contracting, indicating that the expanding clone was dominant. For similar VAF levels at 3 years, participants with expanding CHIP clones had a higher risk of blood cancer than a stable or contracting one (Figure 2). Conclusions: This is the largest longitudinal study of CHIP variants and associated blood cancer outcomes to date. In contrast to other longitudinal studies, there was little evidence for selective growth advantage of CHIP clones with a VAF ≥ 2%, including those with pathogenic mutations of DNMT3A, SF3B1, U2AF1, TP53 and JAK2. While clone size (VAF > 10%) and gene mutation (SF3B1, U2AF1 and TET2) were associated with increased risk of blood cancer progression, our study cautions clinical intervention of CHIP based upon a single timepoint analysis irrespective of the gene mutation. Instead, we propose monitoring of clone size may be a clinically more useful strategy to inform counselling and patient selection for early intervention studies. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

The Combination of Navitoclax and Ruxolitinib in JAK Inhibitor-Naïve Patients with Myelofibrosis Mediates Responses Suggestive of Disease Modification

Background: Myelofibrosis (MF) is a myeloproliferative neoplasm associated with splenomegaly, high symptom burden, bone marrow fibrosis (BMF), and poor prognosis. The Janus kinase inhibitor (JAKi) ruxolitinib is the current standard of care for patients with MF. It improves splenomegaly and disease symptoms with limited impact on disease biology, highlighting an unmet medical need in MF. Driver mutations of JAK2, CALR, and MPL are key to the pathogenesis of MF. Reduction in variant allele frequency (VAF) of driver mutations and reversal of BMF are suggestive of disease modification (Pemmaraju, Cancer, 2022). Additionally, mutations in ASXL1, EZH2, SRSF2, IDH1/2, and U2AF1 p.Q157 are classified as high molecular risk (HMR) mutations associated with a worse prognosis (Arber et al., Blood, 2016). We previously demonstrated evidence of disease modification independent of HMR mutations by adding navitoclax to ruxolitinib in patients with suboptimal responses to ruxolitinib monotherapy (Pemmaraju, Lancet Oncology, 2022). Herein, we investigated whether the type of MF and risk (e.g., age, Dynamic International Prognostic Scoring System [DIPSS], and HMR) correlated with clinical outcomes and responses suggestive of disease modification (e.g., VAF reduction and improvement in BMF) among JAKi treatment-naïve patients with MF treated with the combination of navitoclax and ruxolitinib. Methods: Cohort-3 of the phase-2 multicenter REFINE study (NCT03222609) enrolled JAKi treatment-naïve patients with MF. Enrolled patients had primary or secondary MF with splenomegaly (DIPSS ≥ Intermediate-1) and did not receive prior JAK-2 therapy or bromodomain and extra terminal motif (BET) inhibitors. Patients initiated navitoclax at 100 mg QD or 200 mg QD if baseline platelet count was ≤ 150 × 109/L or > 150 × 109/L, respectively. Ruxolitinib was given BID with starting dose based on baseline platelet count per local label. The primary endpoint was spleen volume reduction of ≥ 35% (SVR35) from baseline at week 24, assessed by MRI conducted by central review. Key secondary and exploratory endpoints evaluated in this analysis were a reduction in BMF obtained from BM biopsies by local evaluation and a reduction in VAF for the driver gene mutations (JAK2V617, CALR, or MPL), respectively. Driver gene VAF and HMR mutations were determined in whole blood with the 50-gene Focus Myeloid next-generation sequencing panel (Interspace Pharma Solutions, Morrisville, NC, USA). Results: As of Feb 07, 2022, all enrolled patients (N = 32) treated with navitoclax and ruxolitinib were evaluable for biomarker analysis. The median duration of follow-up was 6.1 months (range, 1.9 – 18.6); median age was 69 years (range, 44 – 83), and median spleen volume was 1889.08 cm3 (range, 645.6 – 7339.6). At baseline, 22 (69%) patients had JAK2, 6 (19%) had CALR, and 3 (9%) had MPL mutations. One (3%) patient was triple negative. Nineteen (59%) patients had HMR mutations. SVR35 at week 24 was observed in all subgroups known to confer poor prognosis, including age (≥75 years, 50% [n = 4/8]), high DIPSS score (Intermediate-2, 63% [n = 12/19]; high, 33% [n = 1/3]), and HMR mutations (47% [n = 9/19]; Figure). BMF grade improvement was evaluable in 26/32 (81%) patients, of which 9/26 (35%) achieved ≥ 1-grade improvement at any time during treatment with a median time-to-improvement of 12.3 weeks (range, 12.1 – 24.1). Complete resolution of BMF was observed in 2/9 (22%) patients. The baseline BMF grades of these patients were 2 and 3. BMF ≥ 1-grade improvement was achieved by 7/13 (54%) patients with HMR mutations and 2/13 (15%) without HMR mutations. Reduction in driver gene JAK2V617 mutation VAF > 20% from baseline at week 12 or 24 was observed in 14/28 (50%) patients, and 5/14 (36%) patients achieved > 50% VAF reduction from baseline. There were no differences in > 20% VAF reductions from baseline to week 12 or 24 between those with or without HMR mutations (7/15 [47%] versus 7/13 [54%], respectively). Conclusions: Among JAKi treatment-naïve patients with MF, the combination of navitoclax and ruxolitinib reduced splenomegaly in several high-risk groups known to confer poor prognosis. Reductions in BMF and VAF were independent of HMR mutations. The reduction in BMF and VAF for the driver mutation JAK2V617 is encouraging and suggestive of evidence of disease modification with the combination of navitoclax and ruxolitinib. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Myeloid Neoplasms with MYC-Positive Double Minutes, a Specific Subgroup?

Background: Double minutes (dmin) are a rare cytogenetic finding in hematologic malignancies, mostly associated with acute myeloid leukemias. The acentric chromosomal structures usually represent the cytogenetic equivalent of the amplification of an oncogene, such as MYC. Due to their small size dmin are difficult to detect in chromosome banding analysis (CBA) and FISH analysis is required to identify the corresponding gene amplification. In literature several case reports described an association of the presence of MYC-positive dmin with an APL-like cytomorphology, however, larger studies comprehensively characterizing these cases are missing so far. Aim: Characterization of 68 cases of myeloid neoplasms harboring MYC dmin with respect to cytogenetics, morphology, gene expression (GE) and gene mutations. Patients and methods: The study enrolled 68 patients (pts) with myeloid neoplasia and MYC dmin. The diagnosis was established following WHO guidelines. Dmin and MYC amplification were assessed via CBA and FISH. Cases were screened for morphological features indicating a similarity to APL and/or suspected diagnosis of APL (APL-like). NGS of a 72 gene panel and WTS (50 Mio reads) were performed on NovaSeq instruments. Variant calling was performed with Pisces (minimum sensitivity: 3 %). Results: The majority of pts were diagnosed with AML (52/68, 77 %, 7 of these secondary AML) or MDS/AML (3/68, 4 %). Other diagnoses comprised CMML (4/68, 6 %), MPN in acceleration or blast crisis, MDS/MPN overlap or MDS (3/68, 4 % each). APL-like cases amounted to 17/68 (25 %). Auer rods were detected in 24/68 pts (35 %) with a higher frequency in APL-like cases (p=0.007). Myeloperoxidase level was >75 % in 46/54 pts (85 %). CBA revealed a complex karyotype (ck, ≥3 aberrations in addition to MYC dmin) in 21/68 pts (31 %), whereas 19/68 pts (28 %) presented with MYC dmin as the only cytogenetic aberration (MYC dmin only). Most common additional chromosomal gains were +4 (11/68, 16 %), +6 (9/68, 13 %) and +22 (8/68, 12 %), while most frequent chromosomal losses comprised 5q deletions (19/68, 28 %), 17p deletions (18/68, 27 %) and 9q deletions or -9 (16/68, 24 %). Gene mutations (mut) most frequently occurred in TET2 (48/68, 71 %), TP53 (22/67, 33 %), U2AF1 (17/67, 25 %) and NRAS (12/66, 18 %). Almost all pts (66/68, 97 %) harbored at least one mutation in TET2, TP53 or U2AF1. TP53 mut were associated with ck (p<0.001) and were not detected in MYC dmin only pts (p<0.001). EZH2 mut were exclusively observed in cases with non-ck and 5/7 mut occurred in MYC dmin only pts (p=0.014). TET2 mut were observed more frequently in cases with non-ck (89 % vs. 29 %, p<0.001). Similarly, U2AF1 mut co-occurred with ck only in one case (p=0.013). In addition, U2AF1 showed significantly higher mutation frequency in APL-like cases (47 % vs. 18 %, p=0.025). A comparison of the mutation frequencies of all MYC dmin cases to a cohort of AML from routine diagnostics (707 pts) revealed overrepresentation of mut in TP53 (33 % vs. 11 %, p=0.003), TET2 (71 % vs. 19 %, p<0.001), U2AF1 (25 % vs. 4 %, p<0.001) and EZH2 (11 % vs. 2 %, p<0.001) in the cohort of MYC dmin cases, whereas CEBPA, NPM1, FLT3, FLT3-ITD and WT1 mut were only detected in the AML cohort (10 %, p=0.004; 22 %, p<0.001; 8 %, p=0.011, 16 %, p<0.001 %; 8 %, p=0.017 respectively) and IDH2 mut occurred less frequently in MYC dmin cases (2 % vs. 9 %, p=0.022). GE analysis of 40 MYC dmin pts revealed high expression of MPO compared to various AML subtypes, similar to APL (Figure (A)). A tSNE plot based on the top 5 % most variable genes demonstrated similar GE profiles within the MYC dmin cases compared to other AML cases (857 pts) (Figure (B)). A comparison of GE profiles within the MYC dmin cohort revealed differential GE between APL-like and all other MYC dmin cases (46 differentially expressed genes (DEG), adjusted p-value <0.05), as well as between ck and non-ck MYC dmin cases (106 DEG, adjusted p-value <0.05). Conclusions:MYC dmin cases often show an APL-like cytomorphology, Auer rods and high levels of MPO expression. They are characterized by a high frequency of TET2, TP53 and U2AF1 mut as well as a relatively uniform GE profile. Mutation patterns and GE profiles allow further differentiation within MYC dmin cases. In cases with suspected APL after exclusion of a PML::RARA or other RARA fusions CBA and FISH for the detection of MYC dmin is helpful to identify this rare myeloid subtype which is described to be associated with an unfavorable prognosis. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

A First-in-Human Phase I Study of CTX-712 in Patients with Advanced, Relapsed or Refractory Malignant Tumors - Hematologic Malignancies Dose Escalation Cohort

Background: Mutations of RNA splicing factors, including SF3B1, SRSF2, U2AF1, and ZRSR2, are implicated in the pathogenesis of various cancers, such as acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS), most likely via abnormal RNA splicing. CTX-712 is a first-in-class, orally available, highly potent, and selective small-molecular inhibitor of CDC2-like kinase (CLK), a key regulator of the RNA splicing process. Preclinical data show that CTX-712 demonstrated anti-proliferative activity on various in vitro and in vivo models derived from not only solid tumors but also hematologic malignancies. Based on these results, a phase I study of CTX-712 was conducted to determine the recommended dose (RD) in solid tumors and hematologic malignancies. While therapeutic agents have been developed to target RNA splicing factor abnormalities, no such agents have been developed to be standard care until now. We report the interim data from the dose escalation cohort of relapsed or refractory AML and MDS, including preliminary efficacy results of CTX-712. Methods: The objectives of this cohort are to evaluate the maximum tolerated dose, dose- limiting toxicity (DLT), safety, pharmacokinetic/pharmacodynamic profiles, and preliminary efficacy of CTX-712 and to determine the RD for the next phase in patients with hematologic malignancies. A 3+3 design is used for finding the RD. An initial CTX-712 dose of 105 mg twice a week was used based on the safety information of the preceding solid tumor dose escalation cohort. Results: As of April 28, 2022, a total of 6 patients (5 AML and 1 MDS) were enrolled in the initial 105 mg twice-a-week dosing cohort. Three of the 6 patients were considered non-evaluable for DLT analysis (2 patients received less than 75% of the dose in cycle 1 due to tumor lysis syndrome, 1 patient discontinued treatment due to disease progression before completing cycle 1). A DLT of Grade 4 pneumonia was observed in 1 of 3 DLT-evaluable patients. Based on these safety profiles, the safety monitoring committee recommended to suspend the 105 mg twice-a-week dosing cohort and to initiate a 70 mg twice-a-week dosing cohort. Among the safety analysis population (N=6), with a median treatment duration of 114 days (Range 15-609 days), the common any-grade adverse events (AEs) (≥30%) were nausea (100%), diarrhea, hypotension (66.7%), vomiting, blood creatinine increased, pyrexia, insomnia (50.0%), fatigue, malaise, amylase increased, tumour lysis syndrome, dysgeusia, nail discoloration, dry skin, rash, and oedema (33.3%). The most common Grade 3 or higher AE was tumour lysis syndrome (33.3%). In the AML patients (N=5), complete remission (CR) was observed in two patients (40.0%), and CR with incomplete hematologic recovery (CRi) was observed in one patient (20.0%). The median time to remission is 56 days (Range 29-57 days), and the median duration of response is 56 days (Range 49+-532 days). In the MDS patient (N=1), CR was observed. Pharmacokinetic/pharmacodynamic profiles of the hematologic malignancies cohort were comparable with those of the solid tumor cohort in the same dosage. Conclusions: Assessment of CTX-712 safety profile for hematologic malignancies is ongoing. CTX-712 showed preliminary anti-tumor efficacy for both AML and MDS. Clinical trial information: JapicCTI-184188

Incorporation of Mutations and Bone Marrow Fibrosis into the Revised International Prognostic Scoring System in Patients with Myelodysplastic Syndromes

Incorporation of mutations in the revised international prognostic scoring system (IPSS-RM) can improve risk stratification in myelodysplastic syndromes (MDS). The independent prognostic significance of bone marrow fibrosis (BMF) is challenged under currently molecular prognostic models. In this study, clinical and mutational data from 438 MDS patients were analyzed retrospectively. Patients were divided into training (n=306) and validation (n=132) cohorts by randomization. BMF was associated with worse overall survival both in the training cohort and validation cohort. In the training cohort, age, IPSS-R, BMF and mutations in RUNX1, TP53, U2AF1, and BCOR were significantly associated with OS. Independent significant prognostic factors included age, IPSS-R, BMF, TP53 and U2AF1 in multivariate analysis. Using the weighted coefficients of each prognostic factor, we developed a simplified prognostic system that added the impact of age, BMF, TP53 and U2AF1 to IPSS-R. Four risk groups were produced: low, intermediate, high and very high with median OS of not reached, 76, 16 and 11 months in the training cohort, and not reached, 60, 19 and 9 months in the validation cohort, respectively. The new model yielded more separated survival curves than IPSS-R model. More than 50% of patients were re-stratified, especially in IPSS-R intermediate group. In addition, Our model had a better C-index (0.61 in training cohort and 0.63 in validation cohort), compared to IPSS-RM model(0.59 and 0.58) and IPSS-R (0.57 and 0.56). In conclusion, BMF was an independent significant prognostic factor in MDS and adding BMF into prognostic scoring system improved its predictive capacity. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

The Adverse Effect of Myelodysplasia-Related Mutations in De Novo Acute Myeloid Leukemia Is Associated with Higher Variant Allelic Frequency (VAF): A Proposal for a Numeric Cutoff for Variant Allelic Frequency

Mutations in myelodysplasia-related genes ASXL1, BCOR, EHZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1 and ZRSR2 (MRmut) are class-defining alterations in acute myeloid leukemia (AML) in the latest World Health Organization (WHO, with the exception of RUNX1) and International Consensus Classification (ICC) schemes and indicate adverse-risk per European Leukemia Net (ELN) 2022 guidelines, unless occurring in favorable-risk AML. The lower limit of detection (LOD) for most clinical-grade next generation sequencing (NGS) is ~1-2% variant allelic frequency (VAF); however, data on the prognostic significance of MRmut relative to VAF is lacking. No guidelines that provide a numeric cutoff for VAF are available for utilizing these mutations to classify AML. Our aim was to define a clinically meaningful VAF cutoff for MR mutations in de novo AML. We included patients (pts) with de novo, untreated AML with MRmut treated at our institution between 2017 to 2022. We excluded pts with acute promyelocytic leukemia, core binding factor AML, and other ELN adverse-risk categories (adverse karyotype and/or TP53 mutation). NGS at baseline was performed using a panel of 81 genes recurrently mutated in myeloid neoplasms (LOD ~1-2%). We performed a best fit classifier on individual MRmut to determine optimal VAF cutoffs using event-free survival (EFS) and overall survival (OS) as endpoints. VAF for MRmut involving X-linked genes was adjusted for men. Pts (n=145) included 92 (63%) men and 53 (37%) women with a median age of 68 years (range, 58-75). 85 (59%) pts had diploid karyotype AML, 51 (35%) had other intermediate-risk non-diploid karyotype AML, and 9(6%) had insufficient metaphases for analysis. 34 (23%) pts had ELN favorable-risk and 111(77%) had ENL adverse-risk AML. Pts were treated with low-intensity+venetoclax (VEN) (72/50%), low-intensity (26/18%), high-intensity (22/15%), and high-intensity+VEN (12/17%) regimens. Median follow-up was 26.9 months (range, 13.9 - 40.8). 72 (49%) pts had 1 MRmut; the remaining had 2-4 concurrent MRmut. MRmut had the following frequencies and VAFs: ASXL1 (28%; median VAF: 26%, [range 2-48%]); BCOR (18%; median VAF 37% [3-100%]); EHZH2 (9%; median VAF 7% [1-55%]), RUNX1 (32%; median VAF 47% [1-96%]), SF3B1 (9%; median VAF: 38% [1-55%]), SRSF2 (46%; median VAF: 37% [1-76%]), STAG2 (20%; median VAF: 44% [15-98%]), U2AF1 (12%; median VAF: 33% [2-43%]) and ZRSR2 (2%; median VAF: 94% [93-95%]). 34 (23%) pts had ELN favorable-risk AML including 28 (23%) with mutated NPM1 without FLT3-ITD and 6 (4%) with CEBPA bZIP mutations, respectively; in 12/34 pts NPM1 or CEBPA bZIP mutations were subclonal compared with clonal MR mutation(s). A summary of baseline genetic and treatment data is provided in Fig 1A. The overall composite remission rate (CR+CRi) was 88% (64% and 24%, respectively) Median OS was 30.3 months and estimated 12/24 month OS were 77%/60%. The median EFS was 24.6 months and estimated 12/24 month EFS were 65%/51%. Pts with ELN favorable-risk AML with MRmut had longer OS and EFS compared with the adverse-risk group (median OS 34.9 vs 27.5 months, median EFS 34.9 vs 22.9 months) but the difference was not significant (p=0.60 and 0.30, respectively). The VAF cutoff for MRmut associated with EFS and OS, respectively, was: ASXL1: 21%/21%, BCOR 25%/25% , EHZH2 15%/15%, RUNX1 25%/27%, SF3B1 31%/31%, SRSF2 32%/35%, STAG2 40%/21%, U2AF1 44%/44% and ZRSR2 47/47%.There was no significant difference in OS or EFS between pts with 1 MRmut vs those involving ≥2 MR genes. Pts treated with high-intensity, high intensity+ VEN and low intensity+VEN regimens had longer OS (Fig 1B) and EFS compared with pts treated with low intensity regimens without VEN (OS: not reached, 31.7 months, 31.4 months vs 8.9 months, respectively, p<0.001; EFS: not reached, 31.7 months, 29.6 months vs 3.7 months, respectively, p<0.001). Our data suggest that the poor prognostic effect of MRmut in de novo AML is associated with higher VAF and not a binary presence or absence of MRmut. The minimum VAFs associated with worse outcome for each MRmut ranged from 15-47%. Additional analysis to create predictive models is in progress to determine the optimal VAF cutoff when using MRmut to assign AML to the WHO/ICC MR categories and to ELN adverse-risk group. Importantly, our results suggest that pts with MRmut and non-adverse karyotype have significantly better outcomes when treated with high-intensity and/or venetoclax-based treatment regimens. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Molecular Characterization of Response/Loss-of-Response to Ruxolitinib in Patients with Myelofibrosis

Background. Ruxolitinib (RUX) is standard of care for most patients (pts) with primary (PMF) and secondary (sMF) myelofibrosis. However, about half of responders lose benefits at 3-5y, associated with clonal progression, and have a dismal outcome thereafter. The molecular bases for response/loss of response to RUX have not been clarified yet. Aim. To explore genetic variables associated with response/loss of response to RUX in MF pts. Patients and Methods. All consecutive pts with WHO2016/IWG-MRT diagnosis of PMF/sMF, treated with RUX at our Center, were included. To the purpose of the study, response was stringently defined as achievement of spleen lenght reduction from LCM of >50% at any time point; non-responders (NR) were all the others. Loss-of-response was defined as the reappearance of >5cm splenomegaly, if the nadir under RUX was <5cm, and >50% increase if the nadir was >5cm. A custom NGS panel including complete CDS of 552 genes, selected for putative involvement in RUX response (belonging to JAK/STAT, MAPK, ERK pathways, inflammatory reactions, apoptosis) or associated with myeloid neoplasms, was designed. For RNAseq analysis, we used the R package DOSE and performed Over Representation Analysis (ORA) to identify over-represented biological functions/processes. Results. Baseline characteristics. A total of 171 pts were included, 52.1% PMF (15.8% prePMF, 36.5% overtPMF) and 82 sMF (46.9%; 29.2% PPV-, 17.7% PET-MF). DIPSS at RUX start was Int-1: 43.8%, Int-2: 39.8%, high: 16.4%. Median FU from RUX start was 3.5y (0.8-17.3y), median time on RUX was 2.4y (0.2-12y). Sixty-eight pts (39.8%) died, 39 (22.8%) while on RUX. AML developed in 13 pts (7.6%), 10 of which while on RUX. OS from RUX start was 5.8y (4.4-7.3y). JAK2V617F mut in 72.3%, MPLW515 7.6%, CALR 19.3% (12.8% Type1/like, 6.4% Type2/like), TN 2.3%. 68 pts (39.8%) were High Molecular Risk (HMR; ASXL1, EZH2, IDH1, IDH2, SRSF2, U2AF1). Response to RUX. 93/171 pts (54.4%) were NR. Of the 78 responders (45.6%), 45 (57.7%) lost response (hence, R/L) after a median of 22.7 mo from RUX start, and 33 pts (43.3%) maintained response (hence, R/M) at last FU. Response outcome was influenced by RUX dose: 75.2% of NR and 64.4% of R/L had received <20 mg/die, compared with 63.6% of R/M who had received >30 mg/die. RUX stop (any reason) occurred in 24.2% of R/M, 62.2% R/L, 55% of NR (P=0.02). Baseline molecular predictors. Baseline molecular variables found to correlate with response/loss of response included: (i) HMR status, found in 51.1% R/L vs 24.2% R/M and 40.2% NR (P=0.04); (ii) >1 RAS pathway mutated gene (CBL, NRAS, KRAS, PTPN11), found in 30% NR, 5% R/L and 13.8% R/M (P=0.016); (iii) isolated CBL (n=8, 10.5%; P=0.02) and U2AF1 (n=5,6.7%; P=0.05) mutation, found exclusively in NR pts. Nor type nor variant allele frequency (VAF) of driver mutations impacted on response/loss of response. Serial mutational analysis. Among JAK2V617F mut pts, a partial molecular response (PMR, >50% VAF reduction; IWG-MRT criteria) was achieved by 18.6% of pts, after a median of 1.7y (1-5y). Median VAF reduction under RUX of pts with PMR was 71.3% (51-100%). A PMR occurred in 44.4% of R/M vs 14.8% of (R/L+NR), P=0.03. In 14 responder pts who were on stable dose of RUX since >1y and later became R/L, target NGS sequencing for 552 genes was performed using granulocyte DNA, collected at baseline (T0), at the time (+6mo) of response adjudication (T1) and at R/L (T2). A total of 91 mutated genes were acquired at T2 (mean gene number/pt, 2.5, range 1.0-16). Of these, 22 genes belonged to categories of "transcription regulator", 19 "cytokine mediator signaling", 14 "spliceasome/RNA processing", 11 "EGFR/TKI resistance" and 8 "chromatin regulators" (GO, KEGG, Reactome, Wikipathway). Serial, single cell analysis to interpret clonal hierarchy is ongoing, results will be presented at meeting. Pairwise RNAseq was performed in 4 pts with purified PB CD34+ cells collected at T1 and T2. By using a stringent FDR<0.1 value, enriched transcripts belonged to "RHO-GTPase" (n=88), "spliceasome" (n=40), "Type-2 interferon responses" (n=20), "IL12 stimulated gene and proteins" (n=14) categories. Conclusions. This study adds to the understanding of molecular mechanisms of response/loss of response to RUX by identifying baseline predictors and highlighting a set of involved genes/pathways, that remain to be validated.

Prognostic Value of Genomic Clusters Using Machine Learning in Older Adults with AML

Background: Older adults with AML have a decline in performance statues and acquired comorbidities compared to younger patients. Treatment intensity and outcomes were varied according to those factors. Various studies tried to classify genetic landscapes to predict outcomes in AML patients. However, those were mostly for younger patients treated with intensive chemotherapy. Machine learning (ML) is useful for classifying large data by calculating each data and subtracting characteristics. Given the limited data of genomic classification for older adults with AML, we analyzed genetic landscapes using ML in the context of treatment intensity. Methods: Unsupervised hierarchical clustering method was used to determine the most likely set of clusters according to cytogenetic/mutation properties at AML diagnosis. Patient selection criterion for this analysis were set as follows; (a) elderly (60 years or older) AML patients who were diagnosed in Seoul St. Mary's Hospital between Jul 2017 and Oct 2021, (b) having available information on chromosome and gene mutation at diagnosis and (c) treated with either intensive chemotherapy (IC), hypomethylating agents (HMA), or HMA plus Venetoclax (HMA/VEN). Patients who received best supportive care only were not included in this study. Results: A total of 279 patients who met inclusion criteria were selected and their median age at diagnosis was 68 years (range 60-). As first-line treatment, 131 patients received IC (47.0%), 76 patients received HMA only (27.2%), and 72 patients were treated with HMA/VEN (25.8%). After a median follow up of 22.5 months, the median overall survival (OS) in the three groups were 22.7, 8.8, and 12.0 months, respectively (p<0.001). We applied the ELN2022 risk stratification to these 3 different treatment groups. For IC group, there was no survival difference between favorable, intermediate, and adverse groups (44 months, 23.1 months, 14.9 months p=0.255). In addition, HMA groups (p-value =0.903) and HMA/VEN group (p-value = 0.805) did not showed survival differences according to risk groups by ELN2022. Next, we conducted hierarchical clustering analysis in a total cohort, which was found to be best classified into 8 clusters by indices for determining the number of clusters. The dominant cytogenetic features of each cluster are demonstrated as follows. The cluster 1(C1) was a group that IDH1 and TP53 mutations were mostly proportioned; C2 was RNA splicing gene dominant (U2AF1 or SRFS2 mutations); C3 was ASXL1 and complex karyotype dominant; C4 was DNA methylation dominant (DNMT3A or TET2 mutations); C5 was NPM1 dominant; C6 was FLT3 mutations dominant; C7 was RUNX1-RUNX1T1 or CBFB-MYH11 fusion dominant; and C8 was abnormal karyotype [-5 or del(5q); -7;-17/abn(17p)] dominant. Thereafter, we linked these genomic clusters with survival outcomes of each treatment arm. Among the 131 patients in the IC group, C5 and C7 were identified as favorable groups while C2 and C8 were classified as adverse groups. On the other hand, in the HMA group, C3 and C5 were allocated to favorable groups and C7 and C8 as adverse. In the HMA/VEN group, C2 and C6 showed favorable survival outcomes while C1 and C7 showed poorer outcomes. Interestingly, we found that an allocation of each genomic cluster to three distinguished (favorable, intermediate, adverse) prognostic groups was changed by the treatment types and survival outcomes of each cluster was substantially different according to different treatment arms. The groups of C1, C5, and C7 which included the HMA/VEN adverse group showed no differences in OS between HMA and HMA/VEN groups, while IC showed better outcomes (HMA vs. HMA/VEN vs. IC, 9.6 vs 9.2 vs 40.5 months, p<0.001). Among the groups of C2, C4, C6, and C8, the patients showed similar median OS between intensive chemotherapy and HMA/VEN, but both treatments showed better outcomes than HMA only (HMA vs HMA/VEN vs IC, 5.5 vs 17.2 vs 17.4 months, p<0.001). In addition, groups C2 and C4 had better median survival by DAC/VEN therapy than other kinds of treatment. (HMA vs HMA/VEN vs IC, 8.2 vs 17.5 vs Not reached, p=0.039). Conclusion: This study suggests that genomic clustering by unsupervised ML could identify genomic groups by co-occurrence patterns having different survival outcomes according to each treatment modality, thus potentially guiding treatment selection in older adults with AML. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Mutational Landscape, Clonal Evolution and Prognostic Effect of Non-Driver Mutations during Disease Progression of Primary Myelofibrosis

Introduction: Primary myelofibrosis(PMF), including pre-fibrotic myelofibrosis(Pre-PMF) and overt-fibrotic myelofibrosis(Overt-PMF), may finally transform to the terminal stage, which was defined as primary myelofibrosis-accelerate phase/blast phase(PMF-AP/BP).In the past decade, high-throughput next-generation sequencing(NGS) technologies have been applied to identify a large range of genetic mutations in PMF. However, little is known about the difference in mutational landscape and enrichment of non-driver mutations between Pre-PMF, Overt-PMF and PMF-AP/BP patients, as well as the clonal evolution during disease progression of PMF. Therefore, we aimed to clarify the mutational landscape between Pre-PMF, Overt-PMF and PMF-AP/BP patients as well as the clonal evolution and prognostic effect of non-driver mutations during disease progression. Methods: This retrospective cohort included 69 Pre-PMF patients,145 Overt-PMF patients and 45 PMF-AP/BP patients from the Institute of Hematology and Blood Diseases Hospital in China from 2015 to 2021. The targeted panel that covered all exons of 137 genes involved in myeloid malignancies or previously described in MPN was used in patients to detect mutations. X-tile software was used to define a cut-off size of variant allele fractions (VAF) of mutation. ASXL1 mutation VAF≤7.5% was defined as a smaller clone size, while VAF>7.5% was defined as a larger clone size. Results: A total of 259 patients with Pre-PMF, Overt-PMF and PMF-AP/BP were included in this analysis. Driver mutation distribution was JAK2V617F 55.6%, CALRExon9 24.7%, MPLW515 3.9%,triple-negative 15.8%. Most frequent non-driver mutations were ASXL1 28.6%, TET2 18.9%, SRSF2 11.2%, U2AF1 10.1% and SETBP1 7.7%(Figure 1A-C). First, we compared the proportion of non-driver mutations between Pre-PMF, Overt-PMF and PMF-AP/BP. Mutations in the ASXL1(30.3%VS10.1%, P=0.001)and U2AF1 (15.2%VS1.4%, P=0.002)genes were more frequent in Overt-PMF compared with Pre-PMF, whereas mutation in the ARID1A (9.1% VS 0.7%, P=0.005)gene was more frequent in Pre-PMF(Figure1D); as mutations in ASXL1(51.1%VS30.3%,P=0.011), SRSF2(26.7%VS9.7%,P=0.004), RUNX1(22.2%VS5.5%,P=0.001), SETBP1(17.8%VS6.2%,P=0.018), NOTCH2(10.8%VS2.2%,P=0.038), NRAS(15.6%VS3.4%,P=0.004)and EZH2(13.3% VS 4.8%,P=0.048) genes were significantly more frequent in PMF-AP/BP compared with Overt-PMF(Figure 1E). Then, we analyzed the enrichments of non-driver mutations in different subtypes of PMF. ASXL1(OR=4.05,95%CI 1.59～10.31; P=0.003) and U2AF1(OR=12.16,95%CI 1.59～92.22; P=0.002) mutations were enriched in Overt-PMF compared with Pre-PMF, RUNX1 (OR=6.56,95%CI 1.84～23.40; P=0.004)and ASXL1(OR=2.94,95%CI 1.27～6.84; P=0.012) mutations were enriched in PMF-AP/BP compared with Overt-PMF(Figure 1F). Next, we explored clonal evolution during disease progression in 17 paired sequential samples. Non-driver mutations, especially ASXL1(Figure 2A)and RUNX1(Figure 2B)mutations tended to increase their clone sizes and were more likely to be newly acquired than lost during the transformation from the Overt-PMF phase to accelerate/blast phase. ASXL1 mutation was regarded as an adverse prognostic factor in PMF patients according to previous study. However, in Overt-PMF(Figure 2C) and PMF-AP/BP(Figure 2D) patients, ASXL1 mutation had an insignificant effect on overall survival (OS), so we examined the effect of clone size of ASXL1 mutation on OS. Among Overt-PMF patients with ASXL1 mutation, those with smaller clone size had a significantly shorter OS than those with larger clone size(HR=3.94,95%CI:1.09∽14.25; p=0.024) and those without ASXL1 mutation(HR=4.04,95%CI:1.35∽12.11; p=0.007)(Fig 2E). similarly, Among PMF-AP/BP patients with ASXL1 mutation, smaller clone size was associated with significantly inferior OS (HR=4.05,95%CI: 0.89∽18.47;p=0.045) compared with larger clone size (Figure 2F). Conclusion: This study clarified the mutational landscape and enrichments of non-driver mutations between different subtypes of PMF patients. The clonal architecture changes and increment of newly acquired pathogenetic mutations were associated with the disease progression of PMF. Smaller clone size of ASXL1 mutation had an adverse prognostic impact in Overt-PMF and PMF-AP/BP patients. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Molecular and Clinical PHF6 Mutant Myeloid Neoplasia Provides Clues As to Their Pathogenesis and Therapeutic Targeting

Congenital PHF6 mutations (PHF6MT) cause the X-linked neurodevelopmental disorder, Börjeson-Forssman-Lehmann syndrome (BFLS). In contrast, acquired PHF6MT are frequently detected in hematologic neoplasms including acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), and T-cell acute lymphoblastic leukemia (T-ALL). Indeed, PHF6 is highly expressed in brain/developing CNS as well as in hematopoietic subpopulations suggesting its role in neurogenesis and hematopoiesis consistent with the neurologic BFSL presentation of the germline and leukemias with somatic mutations of this gene. This study focuses on involvement of PHF6 lesions in AML. Here, we hypothesize that there are functional links between lymphoid and myeloid neoplasia (MN) with PHF6-deficiency that can be instructive in terms of the impact of this gene as a tumor suppressor gene (TSG). In total, we analyzed sequencing results of 6886 AML cases, including 616 cases from open sources such as The Beat AML Master dataset (Tyner JW , Tognon CE, Bottomly D, et al. Functional genomic landscape of acute myeloid leukaemia. Nature. 2018;562(7728):526-531). We first investigated the clinical phenotype associated with PHF6MT. We detected a total of 111 (1.6% of cases) PHF6MT. While frameshifts were found across all the coding regions, missense and stop-gains were concentrated in the two PHD-type zinc finger domains and were overlapping between other MN and the previously reported T-ALL cases. PHF6MT were significantly more common in male cases (2.3 vs. 0.8%; p<.001) similar to other X-genes such as UBA1, ZRSR2, and STAG2 but different from ATRX, PIGA, and KDM6A (UTX). When comparing primary AML (pAML) with secondary AML (sAML), sAML had more PHF6MT than pAML (2.9 vs. 1.4%; p<.001). While PHF6MT were not associated with chromosomal abnormalities in male cases, female cases with X chromosome deletion (delX) showed higher frequency of PHF6MT compared to females with diploid X chromosome (6.5 vs. 0.8%; p=.0019). Consequently, when delX were included, a total of 2.3% of female patients had PHF6 hits with 10% of them in a homo/hemizygous configuration. While RUNX1, U2AF1, and ASXL1 mutations coincided with PHF6 hits in males, in female cases, delX and ASXL1 mutations were the most common associated defects (Fig.1). Interestingly, RUNX1 mutations (RUNX1MT) in females were enriched in cases with delX. As a result, PHF6MT were enriched into two distinct subclasses: while PHF6MT were enriched in male cases with sAML and RUNX1MT (8.8%), all female cases with sAML, RUNX1MT, and delX also acquired PHF6MT (Fig.2). A higher frequency of PHF6MT in advanced MN than early diseases (e.g., low-risk MDS), suggestive of a secondary rather than a founder role of PHF6MT prompted us to study the clonal architecture of PHF6MT cases using variant allelic frequencies. RUNX1MT were dominant/co-dominant to PHF6 hits in the clonal evolution of AML in contrast to T-ALL, in which PHF6MT appears to be an earlier event. Similar relationship was found with regard to PHF6 hits and other myeloid gene mutations (e.g., ASXL1, DNMT3A, or SRSF2) suggesting a lineage-restricted TSG function of PHF6 in AML following myeloid commitment (consistent with the physiologic expression of PHF6 in pre-B, pre-T, and erythroid progenitors during hematopoietic ontogeny). PHF6 mutations also affected AML patients’ outcome. Overall, PHF6MT showed worse outcome in male cases, but not in female cases. However, if PHF6MT cases also had RUNX1 mutations, the outcome was extremely low in both sexes. Unsupervised RNAseq analysis of PHF6 mutational status vs. expression pattern revealed that PHF6MT cases showed higher mRNA levels of several lymphoid genes (GCSAM, LY9, or BLNK). This finding was also confirmed by RNAseq of in-house PHF6 knock-out K562 and U-937 clones. Immunohistochemistry unveiled that LY9 was positive in illustrative PHF6MT cases and flow cytometry revealed the presence of lymphoid markers in 2/3 PHF6MT cases. Our study confirms that PHF6MT are more common in male AML. Also we demonstrate the intricate relationships between PHF6MT and RUNX1MT in male sAML patients which extend also to female cases with delX. In AML, PHF6MT are a late event compared to T-ALL, in which they occur early in leukemogenesis. This is also in line with higher frequency of PHF6MT in sAML. Some PHF6MT cases have higher expression of lymphoid genes, which may be candidate for molecular targeted therapy. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

U2AF1 Mutant Clonal Hematopoiesis Is Infrequent, Context Dependent, and Commonly Co-Occurs with BCOR Mutations

U2AF1 Mutant Clonal Hematopoiesis Is Infrequent, Context Dependent, and Commonly Co-Occurs with BCOR Mutations

IPSS-M has greater survival predictive accuracy compared with IPSS-R in persons ≥ 60 years with myelodysplastic syndromes

There are considerable new data on mutation topography in persons with myelodysplastic syndromes (MDS). These data have been used to update conventional risk models such as the Revised International Prognostic Scoring System (IPSS-R). Whether the molecular IPSS (IPSS-M) which includes these data improves survival prediction accuracy is untested. To answer this question, we compared survival prediction accuracies of the IPSS-R and IPSS-M in 852 consecutive subjects with de novo MDS. Concordance statistics (C-statistics) of the IPSS-R and IPSS-M in the entire cohort were similar, 0.67 (95% Confidence Interval [CI] 0.64, 0.71) and 0.68 (0.64, 0.71). Average numbers of mutations and of IPSS-M related mutations were greater in persons ≥ 60 years (2.0 [Interquartile Range [IQR], 1, 3] vs. 1.6 [0, 2], P = 0.003; 1.6 [0, 2] vs. 1.3 [0, 2], P = 0.006). Subjects ≥ 60 years had a higher incidence of mutations in RUNX1, TP53, TET2, SRSF2, DNMT3A, STAG2, EZH2 and DDX41. In contrast, mutations in U2AF1 were more common in persons < 60 years. Next we tested survival prediction accuracy based on age < or ≥ 60 years. C-statistics of the IPSS-R and IPSS-M in subjects ≥ 60 years were 0.66 (0.61, 0.71) and 0.69 (0.64, 0.73) whereas in subjects < 60 years they were 0.67 (0.61, 0.72) and 0.65 (0.59, 0.71). These data indicate an advantage for the IPSS-M over the IPSS-R in subjects ≥ 60 years but not in those < 60 years probably because of a great frequency of mutations correlated with survival in those ≥ 60 years.

Effect of mutation allele frequency on the risk stratification of myelodysplastic syndrome patients.

Myelodysplastic syndrome (MDS) is a heterogeneous group of clonal myeloid malignancies. Though several recurrent mutations are closely correlated with clinical outcomes, data concerning the association between mutation variant allele frequencies (VAF) and prognosis are limited. In this study, we performed comprehensive VAF analyses of relevant myeloid-malignancy related mutations in 698 MDS patients and correlated the results with their prognosis. Mutation VAF in DNMT3A, TET2, ASXL1, EZH2, SETBP1, BCOR, SFSF2, ZRSR2, and TP53 mutations correlated with outcomes. In multivariable analysis, DNMT3A and ZRSR2 mutations with high VAF and mutant IDH2, CBL, U2AF1, and TP53 were independent poor prognostic factors for overall survival. A substantial portion of patients in each revised International Prognostic Scoring System (IPSS-R) risk group could be adjusted to different prognostic groups based on the integrated VAF and mutational profiles. Patients with these unfavorable mutations in each IPSS-R risk subgroup had survivals worse than other patients of the same risk but similar to those in the next higher-risk subgroup. Furthermore, patients harboring U2AF1 mutation might benefit from hypomethylating agents. This study demonstrated the critical role of VAF of mutations for risk stratification in MDS patients and may be incorporated in novel scoring systems.