U2AF1 pathogenic variants in myeloid neoplasms and precursor states: distribution of co-mutations and prognostic heterogeneity

We have previously recognized the genotypic and prognostic heterogeneity of U2AF1 mutations (MT) in myelofibrosis (MF) and myelodysplastic syndromes (MDS). In the current study, we considered 179 U2AF1-mutated patients with clonal cytopenia of undetermined significance (CCUS; n = 22), MDS (n = 108), MDS/acute myeloid leukemia (AML; n = 18) and AML (n = 31). U2AF1 variants included S34 (60%), Q157 (35%), and others (5%): corresponding mutational frequencies were 45%, 55%, and 0% in CCUS; 57%, 39%, and 4% in MDS; 61%, 33%, and 6% in MDS/AML; and 55%, 35% and 10% in AML (P = 0.17, 0.36 and 0.09), respectively. Concurrent mutations included ASXL1 (37%), BCOR (19%), RUNX1 (14%), TET2 (15%), DNMT3A (10%), NRAS/KRAS (8%), TP53 (8%), JAK2 (5.5%) and SETBP1 (5%). The two most frequent U2AF1 MT were S34F (n = 97) and Q157P (n = 46); concurrent MT were more likely to be seen with the latter (91% vs 74%; P = 0.01) and abnormal karyotype with the former (70% vs 62%; P = 0.05). U2AF1 S34F MT clustered with BCOR (P = 0.04) and Q157P MT with ASXL1 (P = 0.01) and TP53 (P = 0.03). The median overall survival (OS) in months was significantly worse in AML (14.2) vs MDS/AML (27.3) vs MDS (33.7; P = 0.001); the latter had similar OS with CCUS (30.0). In morphologically high-risk disease (n = 49), defined by ≥10% blood or bone marrow blasts (i.e., AML or MDS/AML), median OS was 14.2 with Q157P vs 37.1 months in the presence of S34F (P = 0.008); transplant-adjusted multivariable analysis confirmed the detrimental impact of Q157P (P = 0.01) on survival and also identified JAK2 MT as an additional risk factor (P = 0.02). OS was favorably affected by allogeneic hematopoietic stem cell transplantation (HR: 0.16, 95% CI; 0.04-0.61, P = 0.007). The current study defines the prevalence and co-mutational profiles of U2AF1 pathogenic variants in AML, MDS/AML, MDS, and CCUS, and suggests prognostic heterogeneity in patients with ≥10% blasts.


INTRODUCTION
Myelodysplastic neoplasm/syndrome (MDS) and acute myeloid leukemia (AML) are heterogeneous diseases, with variable outcomes, largely driven by chromosomal alterations and somatic mutations [1,2].Mutations in genes encoding components of the spliceosome complex (SRSF2, U2AF1, SF3B1, ZRSR2) are observed in approximately one-third of the patients with MDS and nearly half of the patients with MDS transforming to AML (secondary [s] AML) [3][4][5][6].Historically, AML patients with gene mutations related to spliceosome complex, including U2AF1 were considered as intermediate risk, but with recently updated European Leukemia-Net (ELN) 2022 recommendations, these patients are now considered adverse risk based on associated inferior outcomes [7].U2AF1 is an important component of the spliceosome complex required for pre-mRNA splicing [8,9].Mutations in U2AF1 have been described in myeloid neoplasms, with variants causing specific alterations in 3′ splice site recognition [5,10,11].U2AF1 mutations (MT) are typically acquired later in life and associated with rapid rates of progression to MDS and AML [12][13][14][15][16][17] U2AF1 MT predominantly occur at two hotspots (S34F, Q157) located in zinc finger regions [5].An earlier report evaluating clinical outcomes of 78 MDS patients with U2AF1 MT demonstrated that transcriptional factor and epigenetic regulator genes (e.g., ASXL1 [26%], DNMT3A/ PHF6 [12%], BCOR [15%], TET2 [13%], RUNX1/STAG2 [9%], SETBP1[8%]) were predominantly co-mutated in U2AF1 MT myeloid neoplasms.Furthermore, analysis showed that ancestral U2AF1 S34F MT were associated with inferior outcomes in comparison to ancestral U2AF1 Q157 MT, while no differences in outcomes were observed if the mutations were secondary/ subclonal in nature [15].In another study by Tefferi et al. [17], 52 MDS patients with U2AF1 MT were evaluated, S34F and Q157 hotspots were commonly observed.However, a cytogeneticindependent prognostic impact was not evident for either one of the two commonly observed hotspot mutations.Unlike in MDS, the mutational spectrum of U2AF1 MT in myelofibrosis is contrastingly different with higher occurrence of Q157 (50/77 [65%]) compared to S34 (26/77 [34%]) hotspots [10].To the best of our knowledge apart from the aforementioned studies, limited information is available on the molecular profile, myeloid comutation pattern and survival outcomes with unique U2AF1 MT in patients with clonal cytopenia of undetermined significance (CCUS), MDS, MDS/AML and AML.In this report we have analyzed the genomic profile and clinical relevance of U2AF1 MT in a larger cohort of patients with myeloid neoplasms.

Statistical analysis
Continuous variables summarized as medians (range), while categorical variables reported as frequencies (percentage).Unadjusted comparisons of patient characteristics and outcomes among patients with different myeloid neoplasms and U2AF1 MT were made using the Wilcoxon rank sum test (continuous variables) or Fisher's exact test (categorical variables).We derived the cut-offs for U2AF1 VAF by using Receiver Operating Characteristics (ROC) analysis to assess values that correlated with OS.The Kaplan-Meier method was used to estimate overall survival (OS).All tests were two-sided with P value < 0.05 considered statistically significant.Cox proportional hazards regression model was used to determine the univariate and multivariate predictors of overall survival in patients with high-risk myeloid neoplasm.Multivariable models included all significant univariate predictors with P = ≤0.05.We also performed landmark analysis for OS among responding patients from time of response till last follow up or death and evaluated the impact of allogeneic stem cell transplantation (allo-HCT) in these patients.
None of the patients treated in the CCUS group met criteria for an objective response (as adjudicated by MDS response criteria, given that CCUS response criteria do not exist).The complete remission [CR] rates were 25% in the MDS cohort, 18% in MDS/ AML cohort and 26% in AML cohort.Details on responses with regards to diagnostic categories and types of therapies used summarized in Table 1.Overall, 31 (18%) patients underwent allo-HCT; 19 (17.5%) patients with MDS, 4 (22%) patients with MDS/ AML and 8 (26%) patients with AML.Treatment patterns, responses and proportion of patients receiving allo-HCT were not significantly different amongst the two common U2AF1 MT (S34F and Q157P) (Table 2).

Survival outcomes
The median OS of the entire cohort was 26.5 months.The median OS among patients with CCUS, MDS, MDS/AML and AML was 29.2, 33.7, 27.3 and 14.3 months, respectively (P = 0.01; Fig. 2a).The median OS in patients with high-risk myeloid neoplasm was 26.6 months.We then performed a subset analysis among patients with high-risk myeloid neoplasms harboring; S34F and Q157P MT.The median OS were 37.1 vs 14.2 months with U2AF1 S34F and U2AF1 Q157P MT, respectively (P = 0.008; Fig. 2b).The median OS was better in patients with <10% myeloid blasts compared to those with ≥10% myeloid blasts (32.7 vs 21.5 months, P = 0.009; Fig. 2c).We used ROC derived U2AF1 VAF cut off for OS (VAF ≥ 25% vs <25%), however VAF as a continuous variable did not achieve statistical significance (52.3 vs 50.9 months, P = 0.48).Similarly, we also performed landmark analysis for OS among responding patients from time of response till last follow up or death and evaluated outcome with allo-HCT in these patients.The median OS was significantly better with allo-HCT (53.0 months) compared to without allo-HCT (22.8 months), P = 0.04 (Fig. 3).

DISCUSSION
We present data on the molecular profile and survival outcomes of patients with precursor myeloid neoplasms and myeloid neoplasms harboring U2AF1 MT.We observed distinct myeloid comutation profiles and survival outcomes associated with different mutant U2AF1 hotspot regions and amino acid changes.The most frequently observed MT were S34 (60%), Q157 (35%), followed by others (5%), in alignment with prior published data [15].In highrisk myeloid neoplasm patients, U2AF1 Q157P was associated with inferior outcome compared to U2AF1 S34F .U2AF1 Q157P MT significantly clustered with TP53 (15% vs 4%) and ASXL1 (50% vs 28%) mutations and had a higher percentage of co-mutations (91% vs 74%), in comparison to U2AF1 S34F MT.U2AF1 S34F MT on the other hand significantly clustered with BCOR MT (23% vs 0%).Cytogenetic abnormalities were more commonly seen with U2AF1 S34F MT compared to U2AF1 Q157P MT.While the observation that U2AF1 S34F and U2AF1 Q157P are the most frequent U2AF1 MT in myeloid neoplasms is not new [5,[23][24][25], our study is the first to demonstrate their unique co-mutational spectrum and differential prognostic effect.
In our cohort, aligned with most previous observations, we did not observe co-occurrence of other splicing factor MT with U2AF1 MT [26].It is believed that concomitant splicing factor MT in the same cell could be incompatible with survival of the cell.In the new risk stratification schema for MDS and AML, U2AF1 MT are now considered high risk [7,27]; given that 81% of patients in our cohort had concurrent myeloid MT, we asked the question as to whether or not these accompanying MT were acting as confounding factors, adversely influencing U2AF1 MT related outcomes.In our analysis, the OS of the entire cohort including patients with CCUS, MDS, MDS/AML and AML was sub-optimal at 26.5 months, with a differential prognostic impact imparted by the two U2AF1 hotspot regions; 37.1 vs 14.2 months (P = 0.008) in patients with high-risk myeloid neoplasm harboring U2AF1 S34F  and U2AF1 Q157P MT, respectively.Earlier reports suggested differential degree of anemia and thrombocytopenia with different U2AF1 MT.In MDS, thrombocytopenia was specifically associated with U2AF1 S34F MT and anemia with U2AF1 Q157 MT [24].In patients with myelofibrosis, both mutation types were associated with anemia and the association with thrombocytopenia was most evident with U2AF1 Q157 MT [10].In our cohort of patients with CCUS, MDS, MDS/AML and AML, we did not observe significant differences in anemia or thrombocytopenia among different U2AF1 MT.Interestingly, in the current study we observe a relatively higher rate of CG abnormalities and a shorter OS in U2AF1 MT CCUS patients, in comparison to patients with MDS and AML.Our group has recently reported on clinical outcomes of patients with U2AF1 MT clonal hematopoiesis (CHIP [clonal hematopoiesis of indeterminate potential] and CCUS) [16].In that study, we observed a high rate (25%) and a short latency (17.5 months) towards progression to myeloid neoplasms.We acknowledge that the higher incidence of CG abnormalities in the CCUS group could have been due to selection bias, inherent to the structure of retrospective studies.
Current literature suggest sub-optimal responses with hypomethylating agent therapies in U2AF1-mutated myeloid neoplasms [28], and better responses with HMA plus venetoclax-based therapies [6].Similarly, we observed lower CR rates with HMA therapy alone (12%) and relatively better responses with HMA plus venetoclax (45%) or intensive chemotherapies (65%).However, our sample size was small and larger prospective studies are needed to gauge responses with different treatment regimens.Similar to earlier reports, our study suggests benefit from allo-HCT in improving survival outcome among patients with U2AF1 MT myeloid neoplasms [29].
We acknowledge the limitations of our analysis including inherent selection bias and lack of sequential mutation testing in this cohort of patients with myeloid neoplasms.Nevertheless, our study highlights the unique co-mutation patterns and survival outcomes in CCUS, MDS and AML patients with different U2AF1 MT, underscoring the need for accurate mutational assessment and reporting.

Fig. 1
Fig. 1 U2AF1 mutation structure and somatic co-mutation pattern.a Overview of U2AF1 domains, structures, and distribution of U2AF1 mutations detected, positioned on the U2AF1 protein.Protein Sequence of ZF1 (hotspots at codon 34; S34F and S34Y) and ZF2 (hotspots at codon 157; Q157R and Q157P) domains, where all U2AF1 mutations clustered.b Patterns of the co-mutations identified in the U2AF1 cohort with respective mutations.NTD N-terminal domain, ZF zinc finger domain, RRM RNA recognition motifs, RS The C-terminal Arg-Ser rich domain, CTD C-terminal domain.

Fig. 2
Fig. 2 Kaplan-Meier overall survival curves.a In CCUS, MDS, MDS/ AML and AML, b in high-risk myeloid neoplasm with S34F and Q157P mutations and c in relation to bone marrow blast <10% vs ≥10%.

Fig. 3
Fig. 3 Landmark analysis for OS.Landmark analysis for overall survival among responding patients receiving allogeneic stem cell transplantation (allo-HCT).

Table 1 .
[20,21]e characteristics of patients with U2AF1 mutation; N = 179 (%)/[range].We reviewed the Mayo Clinic database of patients with myeloid neoplasms who underwent next-generation genomic sequencing (NGS) between January 2015 and July 2021.We evaluated the molecular profile and outcomes in 179 patients with precursor myeloid neoplasms (clonal cytopenias of undetermined significance [CCUS; n = 22]) and myeloid neoplasms (MDS [n = 108], MDS/AML [n = 18] and AML [n = 31]) harboring U2AF1 MT.Clinical NGS testing was performed on DNA extracted from fresh bone marrow aspirates.The Mayo Clinic NGS panel included 42 genes (Supplementary Material) and has an accuracy of >99% and reproducibility of 100% for single base substitutions and insertion/ deletion events.The panel has a variant sensitivity of ≥2% VAF with a minimum depth coverage of 250x.CCUS was defined according to the 2022 WHO (World Health Organization) criteria; MDS, MDS/AML and AML were defined as per International Consensus Classification (ICC) 2022 of myeloid neoplasms and acute leukemia[18,19].For this analysis, we operationally defined high-risk myeloid neoplasms as myeloid neoplasms with ≥10% myeloid blasts in the peripheral blood and/or bone marrow.Treatment responses in MDS and AML were assessed according to the International Working Group (IWG) MDS response criteria (2006) and the 2017 ELN AML response criteria, respectively[20,21].
included these in the MDS category, as long as they did not meet criteria for morphological dysplasia.PATIENT AND METHODS