Splicing factor mutations clearance and outcomes in clonal myeloid neoplasms: a referral center experience Authors: Manasawee Tanariyakul, Sarah Mott, Bethany Daniels, Avani Yaganti, Chalothorn Wannaphut, Sarah Hornberg, Prajwal Dhakal, Grerk Sutamtewagul, Kittika Poonsombudlert Published: 2026 ABSTRACT Splicing factor (SF) mutations (SF3B1, SRSF2, U2AF1, and ZRSR2) are implicated in the pathogenesis of myelodysplastic syndrome (MDS), chronic myelomonocytic leukemia (CMML), and acute myeloid leukemia (AML). However, additional cytogenetic aberrations or driver mutations are typically required for the development of AML, raising questions about the utility of SF mutation persistence as a marker of measurable residual disease (MRD). We conducted a retrospective review of adult patients treated at UIHC from 2016 to 2025. Logistic and Cox regression models assessed the impact of clinical and molecular factors on composite complete remission (CR), allogeneic stem cell transplantation (HSCT), AML transformation, and overall survival (OS). Among 95 patients (AML, N = 57; MDS/CMML, N = 38) who received treatment of cytoreductive potential, SF mutations became undetectable in 22 AML and 8 MDS/CMML patients. In AML, post-treatment SF mutation status was associated with receipt of HSCT, but not with CR or OS. In MDS/CMML, SF mutation clearance was associated with CR, but not the receipt of HSCT or OS. Median OS was 27.2 vs. 17.2 months in AML and 16.7 vs. 23.7 months in MDS/CMML for clearance versus persistence groups, respectively. These findings suggest that SF mutation clearance does not significantly impact OS but may influence other clinical outcomes in patients with myeloid neoplasms harboring SF mutations. Introduction Measurable residual disease (MRD) is a valuable prognostic tool in hematologic malignancies. Certain MRD assays can guide consolidative treatment decisions [1–4] or even predict the success of treatment-free remission [5]. In clonal myeloid neoplasms (MNs), including acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), and chronic myelomonocytic leukemia (CMML), a universally reliable MRD marker remains an area of active research [6], due to the limitations of currently available modalities such as multicolor flow cytometry (MFC), polymerase chain reaction (PCR), and next-generation sequencing (NGS). Clonal architecture and the timing of mutation acquisition remain key factors in pathogenesis in MNs [4, 7]. Some mutations, such as NPM1 or FLT3-ITD, independently drive leukemogenesis and are considered clinically significant when detected as MRD [8], while others, such as DNMT3A, TET2, or ASXL1 (collectively “DTA”), can persist as background clonal hematopoiesis (CH) after remission has been achieved [9]. Splicing factor (SF) mutations (SF3B1, SRSF2, U2AF1, and ZRSR2) are considered early or founding events in MNs [8], often requiring additional cytogenetic or molecular abnormalities for leukemogenesis, but the prognostic significance of SF mutation persistence on overall survival (OS), achievement of complete remission (CR), and receipt of an allogeneic stem cell transplant (HSCT) following the first line of cytoreductive treatment has not yet been established. Methods We retrospectively reviewed electronic medical records of adult patients with AML, MDS, or CMML treated at the University of Iowa Health Care (UIHC) between 2016 and 2025. Patients provided informed consent for treatment and authorized the use of their personal information for research purposes. This study was approved by the University of Iowa Institutional Review Board. Patient population Patients included in this study were adults aged 18 years and older diagnosed with MDS, CMML, or AML based on WHO 2022 criteria [10]. AML patients were risk-stratified by ELN20229, while MDS and CMML patients were stratified by IPSS-M score [11] and CPSS-Mol [12], respectively. Cytogenetic risks are considered “adverse” when defined as intermediate or worse according to each prognostic model. All patients had one of the SF mutations at the time of diagnosis and subsequent molecular testing after the first line of cytoreductive treatment. Treatments Only patients who received treatment of cytoreductive potential according to their disease risk category and fitness status were included in this study. High-intensity regimens typically included anthracycline/cytarabine (“7 + 3”, or liposomal doxorubicin-cytarabine) or fludarabine/cytarabine/idarubicin/filgrastim (“FLAG-Ida”) while low-intensity regimens included hypomethylating agent (HMA) +/- venetoclax (ven), and consolidative HSCT was pursued for eligible patients. Response assessment Disease assessment included morphology, MFC (limit of detection (LOD) 0.01%), cytogenetics, and mutation test by next-generation sequencing (NGS) (LOD variant allelic frequency (VAF) 4%). PCR was performed for AML patients with NPM1 co-mutation, but due to the rarity in our cohort, a separate analysis was not performed. Response assessment was evaluated according to the ELN 2022 for AML patients and the IMWG 2023 guideline for MDS/CMML patients [9, 13]. In this study, CR was defined as composite CR (CRc), which includes CR, CR with incomplete hematologic response (CRi), and CR with partial hematologic recovery (CRh). Time to first response was at count recovery for patients receiving high-intensity regimens, and after 3–4 cycles for patients who received low-intensity regimens. All HSCT patients were evaluated for pre-HSCT disease status at 2–4 weeks before conditioning chemotherapy and for post-HSCT disease status according to the standard of care. Statistical analysis Logistic regression models were utilized to estimate the effect of patient, disease, and treatment characteristics on the odds of CR after initial treatment and proceeding to HSCT. Survival probabilities were estimated and plotted using the Kaplan-Meier method. Time was calculated from initial treatment to death due to any cause for OS. Otherwise, patients were censored at last contact. Cox regression models were used to estimate the effect of patient, disease, and treatment characteristics on OS. All statistical testing was two-sided and assessed for significance at the 5% level using SAS v9.4 (SAS Institute, Cary, NC). Results Table 1 outlines the patient characteristics by diagnosis. Baseline demographics and co-mutation patterns were balanced across SF mutations.Table 1Patient characteristicsAML (N = 57)MDS/CMML (N = 38)Median age (range)72 (22–93)71 (63–85)Male34 (60%)31 (82%)Median % blast (range)43 (8–100)6 (1–17)Median IPSS-M (range)-0.3 (−1.5–4.2)CytogeneticNormal26 (47%)19 (50%)Other*11 (20%)8 (21%)Adverse18 (33%)11 (29%)Mutation in patients (N)U2AF11010SRSF23618ZRSR245SF3B175Initial mutation median VAF (%)U2AF118%26%SRSF263%47%ZRSR27%13%SF3B112%13%TreatmentChemotherapy18 (32%)0HMA+venetoclax36 (63%)0HMA3 (5%)38 (100%)Transformation to AMLYes-6 (16%)No-32 (84%)HSCT conditioning**11 (19%)15 (40%)RIC6 (55%)15 (100%)MAC5 (45%)0Months of follow-up (range)16.6 (1.6–85.0)14.8 (2.5–92.1)* Trisomy 8 was the most common among cytogenetic changes in this patient group, no patients with core-binding factor AML were included.**The conditioning regimen was classified as MAC when it included fractionated total body irradiation > 8 Gy or intravenous busulfan ≥ 9 mg/kg; otherwise, it was classified as RIC. Table 1Patient characteristicsAML (N = 57)MDS/CMML (N = 38)Median age (range)72 (22–93)71 (63–85)Male34 (60%)31 (82%)Median % blast (range)43 (8–100)6 (1–17)Median IPSS-M (range)-0.3 (−1.5–4.2)CytogeneticNormal26 (47%)19 (50%)Other*11 (20%)8 (21%)Adverse18 (33%)11 (29%)Mutation in patients (N)U2AF11010SRSF23618ZRSR245SF3B175Initial mutation median VAF (%)U2AF118%26%SRSF263%47%ZRSR27%13%SF3B112%13%TreatmentChemotherapy18 (32%)0HMA+venetoclax36 (63%)0HMA3 (5%)38 (100%)Transformation to AMLYes-6 (16%)No-32 (84%)HSCT conditioning**11 (19%)15 (40%)RIC6 (55%)15 (100%)MAC5 (45%)0Months of follow-up (range)16.6 (1.6–85.0)14.8 (2.5–92.1)* Trisomy 8 was the most common among cytogenetic changes in this patient group, no patients with core-binding factor AML were included.**The conditioning regimen was classified as MAC when it included fractionated total body irradiation > 8 Gy or intravenous busulfan ≥ 9 mg/kg; otherwise, it was classified as RIC. Patient characteristics * Trisomy 8 was the most common among cytogenetic changes in this patient group, no patients with core-binding factor AML were included. **The conditioning regimen was classified as MAC when it included fractionated total body irradiation > 8 Gy or intravenous busulfan ≥ 9 mg/kg; otherwise, it was classified as RIC. AML cohort Among AML patients (N = 57; median age 72; 60% male), treatments included chemotherapy (N = 18), HMA/venetoclax (n = 36), and HMA alone (N = 3), with a CR rate of 75%. SF mutation clearance occurred in 22 (29%) patients and was not associated with CR (Odd Ratio (OR): 1.80, 95% Confidence Interval (CI): 0.49–6.66). Persistent SF mutations, older age, and low-intensity therapy were associated with lower odds of receiving an HSCT on univariate analysis. OS did not differ between clearance and persistence groups (median: 27.2 vs. 18.7 months, p = 0.07). (Fig. 1).Fig. 1OS of AML and MDS/MPN among clearance vs. persistence of SF mutations OS of AML and MDS/MPN among clearance vs. persistence of SF mutations On multivariable analysis (MVA), RAS pathway co-mutations (Hazard Ratio (HR): 2.61, 95% CI 1.12–6.06) and not achieving CR (HR: 3.60, 95% CI 1.63–7.99) were associated with decreased OS. (Table 2).Table 2Multivariable analysis of variables associated with OS in AMLOverall Survival----------------------------------------CovariateLevelNHazard Ratio95% CIP-valueSexMale341.560.693.520.29Female23Ref--RASYes102.611.126.060.03No47Ref--CR After Initial TreatmentNo143.601.637.99< 0.01Yes43Ref--*Number of observations in the original data set = 57. Number of observations used = 57 Table 2Multivariable analysis of variables associated with OS in AMLOverall Survival----------------------------------------CovariateLevelNHazard Ratio95% CIP-valueSexMale341.560.693.520.29Female23Ref--RASYes102.611.126.060.03No47Ref--CR After Initial TreatmentNo143.601.637.99< 0.01Yes43Ref--*Number of observations in the original data set = 57. Number of observations used = 57 Multivariable analysis of variables associated with OS in AML *Number of observations in the original data set = 57. Number of observations used = 57 After treatment with chemo vs. HMA +/- ven, we did not find a statistically significant difference in the rate of SF mutation clearance (p = 0.54). OS did not significantly differ based on SF mutation subtypes. (Fig. 2)Fig. 2OS of AML patients stratified by SF mutation OS of AML patients stratified by SF mutation MDS/CMML cohort Among MDS/CMML patients (N = 38; median age 71; 82% male), all received HMA, with a CR rate of 39%. SF mutation clearance occurred in 8 (21%) patients and was associated with CR (OR 7.00, 95% CI 1.18–41.54). (Fig. 1) Six patients transformed to AML despite treatment, and 15 underwent HSCT. On univariate analysis (UVA), increasing age was associated with lower odds of receipt of a HSCT (OR: 0.69, 95% CI 0.54–0.89), whereas SF mutation clearance was not (OR: 0.3, 95% CI: 0.06–1.52). OS did not differ between clearance and persistence groups (median 16.7 vs. 23.7 months, p = 0.19) (Table 3). After treatment with azacitidine vs. decitabine, we did not find a statistically significant difference in the rate of SF mutation clearance (p = 0.29). OS did not significantly differ based on SF mutation subtypes. (Fig. 3).Table 3Univariate analysis for odds of proceeding to HSCTAMLMDS/CMML------------------------------------------------------------------------------------------CovariateLevelOdds Ratio95% CIP-valueOdds95% CIP-valueSexMale0.770.202.910.700.410.082.190.30FemaleRef--Ref--CytogeneticAdverse2.750.6511.690.233.360.7016.080.26Other0.550.055.572.800.5015.66NormalRef--Ref--RASYes0.410.053.640.42NoRef--NPM1Yes2.350.569.800.24NoRef--IDH1/2Yes1.820.457.390.40NoRef--DTAYes0.340.081.460.150.800.222.950.74NoRef--Ref--RUNX1Yes1.350.306.050.69NoRef--Initial TreatmentHMA +/- Ven0.020.000.19< 0.01ChemoRef--Initial TreatmentDecitabine0.560.152.100.39AzacitidineRef--CR After Initial TreatmentNo1.190.275.290.820.140.030.60< 0.01YesRef--Ref--Splicing Mutation StatusPersistent0.160.040.710.020.300.061.520.15ClearanceRef--Ref--Age at DiagnosisUnits = 10.870.790.97< 0.010.690.540.89< 0.01% BlastsUnits = 11.000.971.030.961.060.921.230.39VAF at DiagnosisUnits = 10.980.941.020.330.980.941.020.35IPSS-MUnits = 11.390.872.240.17Fig. 3OS of MDS/CMML patients stratified by SF mutation Table 3Univariate analysis for odds of proceeding to HSCTAMLMDS/CMML------------------------------------------------------------------------------------------CovariateLevelOdds Ratio95% CIP-valueOdds95% CIP-valueSexMale0.770.202.910.700.410.082.190.30FemaleRef--Ref--CytogeneticAdverse2.750.6511.690.233.360.7016.080.26Other0.550.055.572.800.5015.66NormalRef--Ref--RASYes0.410.053.640.42NoRef--NPM1Yes2.350.569.800.24NoRef--IDH1/2Yes1.820.457.390.40NoRef--DTAYes0.340.081.460.150.800.222.950.74NoRef--Ref--RUNX1Yes1.350.306.050.69NoRef--Initial TreatmentHMA +/- Ven0.020.000.19< 0.01ChemoRef--Initial TreatmentDecitabine0.560.152.100.39AzacitidineRef--CR After Initial TreatmentNo1.190.275.290.820.140.030.60< 0.01YesRef--Ref--Splicing Mutation StatusPersistent0.160.040.710.020.300.061.520.15ClearanceRef--Ref--Age at DiagnosisUnits = 10.870.790.97< 0.010.690.540.89< 0.01% BlastsUnits = 11.000.971.030.961.060.921.230.39VAF at DiagnosisUnits = 10.980.941.020.330.980.941.020.35IPSS-MUnits = 11.390.872.240.17 Univariate analysis for odds of proceeding to HSCT OS of MDS/CMML patients stratified by SF mutation Discussion We did not find a statistically significant difference in OS based on persistence vs. clearance of SF mutation after cytoreductive therapy among patients with MNs. However, the effect of SF mutation clearance was more variable when evaluating other related clinical outcomes, such as achievement of CR or receipt of an HSCT. While additional data are needed to confirm our findings, we speculate that SF mutations may rank lower in the clonal hierarchy with attenuated functional impact relative to stronger driver co-mutations or cytogenetic aberrancies. Although a 20% blasts threshold by morphology is traditionally used to distinguish AML vs. other MNs, we recognized that a more recent classification, such as TP53-mutated MDS/AML9 that prioritize driver mutations over blast percentage may better reflect MNs as a disease spectrum rather than distinguished entities. For example, our cohort included three patients with SF3B1-mutated MDS/CMML who initially presented with anemia-predominant disease but later acquired MECOM rearrangement (MECOMr) at progression with refractoriness to cytoreductive treatments, which echoes the findings from Huber et al [14]. regarding the association of SF3B1 and the subsequent evolution of MECOMr. From DNA to protein production, messenger ribonucleic acid (mRNA) and pre-mRNA are important intermediary steps in the generation of functional proteins. The spliceosome complex tightly regulates alternative splicing, enabling the production of diverse protein isoforms. However, the presence of multiple splice sites also creates opportunities for mutations that can alter downstream protein function and contribute to several types of malignancies, including MNs [15–17]. Notably, in our subset analyses of individual SF mutations, we did not detect a significant imbalance in rates of adverse cytogenetics or co-mutation patterns. Prior reports suggest enrichment of adverse risk cytogenetics and TP53 co-mutation among the SRSF2-mutated MDS/CMML/AML [17] as the source of their relatively poor response to conventional cytoreductive therapies [16], but this was not seen in our cohort and our SRSF2-mutated population had similar OS compared to other SF mutations. The SF3B1 mutation is generally associated with a favorable risk in MDS [18], especially among the K700E variants [19]. However, all patients in our cohort required cytoreductive treatment and more commonly had non-K700E SF3B1 variants, which may explain the less favorable outcomes. While no statistically significant difference in OS between SF mutation was detected, we observed worse OS in the AML cohort but not in the MDS/CMML cohort among U2AF1-mutated patients. We did not find a significant effect on the variables in our univariate analysis or after comparing between S34F or Q157R variants, which may be secondary to other undiscovered factors not captured in our analysis. The ZRSR2 mutation was the least prevalent among the SF mutations identified in our cohort, which is concordant with findings from prior studies [20, 21]. A larger, longer-term follow-up is needed to clarify the significance of ZRSR2 mutation, as they are likely heavily influenced by DTA co-mutations [20]. As we learn more about the spliceosome complex, novel pathogenic variants such as PRPF8 and LUC7L2 continue to be discovered, but the clinical significance and proper management approach of these mutations remain unknown [17]. Efforts are underway for the development of new therapeutic options for SF-mutated clonal myeloid neoplasms [22–25]. The strengths of this study include long-term follow-up, comprehensive evaluation of pre- and post-treatment NGS tests, and inclusion of post-HSCT outcomes. While we acknowledge the significant heterogeneity in the natural course of disease among the MN patients, the relatively small sample size did not allow us to evaluate the impact of different SF mutation variants and the clone sizes. A more detailed serial single-cell study [4] may be needed to elucidate the true impact of the timing of SF mutation emergence and clearance in relation to other driver pathways. Conclusion As the assays for MRD become increasingly sensitive, distinguishing true MRD vs. background CH becomes crucial in defining each patient’s remission status. Our data suggest that the clearance of SF mutations does not statistically impact the OS of patients with MNs but may influence other clinical outcomes in those harboring SF mutations. The effect of each SF mutation variant and clone size in relation to other co-mutations/cytogenetic abnormalities deserves further investigation in a large multi-center study.