Background/Objectives: Acute myeloid leukemia (AML) comprises genetic subclasses with distinct gene expression profiles. While AML gene expression studies have mainly focused on protein-coding genes, our understanding of expression patterns of long intergenic noncoding RNAs (lincRNAs) remains incomplete. This is due to limited sample sizes, as well as incomplete annotation of lncRNAs with context-dependent expression. Methods: To address this gap, we developed the bioinformatic pipeline LIRA (long intergenic noncoding RNA annotator) to identify novel lincRNAs using stringent criteria, including spliced and intergenic transcripts, and algorithms to exclude coding potential. Results: By applying LIRA to RNA-sequencing data from 878 pediatric and adult AML cases and 20 healthy controls, we identified 1560 novel lincRNAs, expanding the GENCODE v38 lincRNA catalog by 27%. Integration of in-house-generated CAGE- and ChIP-sequencing data from KMT2A::MLLT3 samples revealed that 80% of the novel lincRNAs are 5' capped, and at least 67% harbor activating epigenetic marks at their transcription start sites. Unsupervised analysis of the 1000 most variable known and newly identified lincRNAs uncovered subclass-specific expression patterns, mirroring those observed for protein-coding genes. Weighted Gene Co-expression Network Analysis identified 17 lincRNA expression modules associated with AML subclasses. Notably, expression of these modules decreased upon degradation of the leukemogenic onco-fusion proteins KMT2A::MLLT3 and PML::RARA, indicating that lincRNA expression is responsive to oncogenic signaling. Conclusions: This comprehensive analysis shows that lincRNAs exhibit similar subclass-specific expression patterns as protein-coding genes and establishes a valuable resource for future studies on genetically defined AML subclasses, with potential implications for biomarker discovery and therapeutic targeting.

While TP53 mutations in myeloproliferative neoplasms (MPN) are associated with an increased risk of leukemic transformation, not all patients carrying a TP53 mutation progress. To better risk-stratify MPN patients with TP53 mutations, we analyzed data from 1540 patients treated at four specialized cancer centers. Among them, 1429 had wildtype TP53 and 111 had mutations in the TP53 gene. At first MPN diagnosis, 32% had polycythemia vera, 39% had essential thrombocythemia, and 25% had primary myelofibrosis. Among all MPN patients with TP53 mutations, presence of fibrosis in the bone marrow (hazard ratio (HR): 3.84, 95% CI: 1.98-7.43), multi-hit TP53 mutation status (HR: 2.74, 95% confidence interval (CI): 1.52-4.97), and higher PHANTM score (HR: 1.87, 95% CI: 1.02-3.42) were associated with worse OS in a multivariable analysis. Based on these variables, we developed a risk model to identify TP53-mutated MPN patients who are at high risk for inferior OS. Median OS from time of TP53 detection was 0.5 years in high-risk patients, compared to 2.3 years for patients with intermediate risk and 6.3 years for patients with low risk. This scoring system may help refine risk stratification for chronic phase MPN patients harboring TP53 aberrations.

Somatic mutations in RNA splicing regulators, including the serine/arginine-rich protein SRSF2, are frequently observed in myeloid malignancies. Using mouse models and primary human samples, we investigated the impact of SRSF2 mutations on erythropoiesis. We found reduced erythropoiesis in Srsf2P95H versus wild-type mice upon stress-induced erythropoiesis and identified that SRSF2 mutations correlate with reduced hemoglobin in JAK2-mutant patients with myeloproliferative neoplasms (MPN). Consistent with this, Jak2V617F-Srsf2P95H versus Jak2V617F mice displayed reduced red blood cell counts and erythroid precursor frequencies. RNA-sequencing on erythroid precursors showed reduced expression of heme metabolism and mitotic spindle-related genes, and increased expression of mTORC1 signaling in Srsf2P95H versus wild-type cells. RNA splicing analyses on the same cells and on human patient samples identified aberrant FYN splicing in SRSF2mut cells, with increased aberrant FYNB over normal FYNT transcripts. FYNB, but not FYNT, expression resulted in reduced erythroid differentiation and increased phosphorylation of mTORC1 downstream target S6. Additionally, increased S6 phosphorylation was confirmed in primary Srsf2P95H erythroid cells. mTORC1 pathway inhibition using rapamycin normalized FYNB- and Srsf2P95H-induced impaired erythropoiesis and significantly increased erythroid colony formation of SRSF2-mutant myelodysplastic neoplasm (MDS) bone marrow cells. Our data reveal targetable molecular mechanisms of impaired erythropoiesis in SRSF2-mutant cells.

Cardiovascular risk factors (CVRFs) are important modifiers of thrombosis in patients with essential thrombocythemia (ET), polycythemia vera (PV), and myelofibrosis (MF). We performed a retrospective cohort analysis evaluating CVRFs in 1005 patients with myeloproliferative neoplasms (MPNs) from the Dana-Farber Cancer Institute Hematologic Malignancies Data Repository from 2014 to 2023. We also included a non-MPN group of 1543 age- and sex-matched controls with no known diagnoses of hematologic malignancies to evaluate whether CVRFs differentially affected outcomes. CVRFs were identified through International Classification of Diseases codes for hypertension, hyperlipidemia, type 2 diabetes mellitus (DM2), current smoking, or body mass index ≥30 before MPN diagnosis. CVRFs occurred in 34% of patients with MPNs. Patients with MPN with ≥1 CVRF had increased risk of death (hazard ratio [HR], 2.52; 95% confidence interval [CI], 1.9-3.35) and arterial/venous thrombosis (HR, 3.05; 95% CI 2.39-3.92). Within MPN subtypes, patients with ET, PV, and MF who had CVRFs also demonstrated worse overall survival and thrombotic outcomes. Among CVRFs, only DM2 predicted worse thrombotic outcomes in patients with MPNs. The HR of CVRF on thrombosis was decreased in patients with MPNs compared with non-MPN controls (HR, 0.51; 95% CI, 0.36-0.86). Looking at ET, PV, and MF specifically, the presence of a CVRF also had less of an impact on thrombotic risk in ET compared with controls (HR, 0.35; P = .019); no interactions between MPN diagnosis and CVRFs were seen in patients with PV and MF. Our results underscore both the necessity of managing CVRFs in MPNs to improve patient morbidity and mortality and the need to ameliorate thrombotic risk with measures beyond addressing CVRFs.

Introduction: Although the inferred “fitness” of clonal hematopoesis (CH) driver mutations differs depending on the gene mutation, the impact of environmental factors that may promote or impair CH expansion remain largely unknown. We hypothesized that obesity influences the clonal expansion rate of common CH driver mutations. Methods: To test this hypothesis, we first interrogated data from the UKBiobank (UKBB) (n=425,573 exomes) to evaluate the relationship between body mass index (BMI) (kg/m2) and four common CH mutations, namely DNMT3A, TET2, ASXL1, and JAK2. For functional studies we focused on JAK2V617F, using a novel Fgd5-CreER-Jak2V617F mouse model we generated. In this model, Jak2V617F expression is induced in a small percentage of long-term (LT)-HSC which are tracked sequentially in primary mice in an unirradiated bone marrow niche. To quantify Jak2V617F expression in the model, we developed a digital droplet PCR assay. We also generated a chimeric bone marrow transplant Jak2V617F model and performed bulk RNA-sequencing (RNAseq) on purified lineagelow Sca-1+ c-kithigh (LSK) cells expressing Jak2V617F or wild-type Jak2, isolated from the same mouse. Finally, we performed single-cell RNAseq on Jak2V617F-expressing c-kithigh cells purified from chimeric transplant mice. Mice fed an obesity diet received 60 kcal% fat while mice fed a control diet received 10 kcal% fat. Results: In UKBB analyses, we identified genotype-specific patterns of association between BMI and the presence of CH mutations. JAK2 (OR 0.51; 95% CI 0.27-0.96; p=0.036) and DNMT3A mutations (OR 0.92; 95% CI 0.87-0.97; p=0.0028) were negatively associated with a BMI >30 relative to BMI <25, whereas mutations in ASXL1(OR 1.40; 95% CI 1.23-1.60; p=8.8x10-7) and TET2 (OR 1.16; 95% CI 1.04-1.31; p=0.0095) were positively associated with BMI >30. To further explore the negative association between JAK2V617F and obesity, we generated a cohort of Fgd5-CreER-Jak2V617F mice and fed half the mice an obesity diet and the other half a control diet. After 24 weeks, Jak2-mutant mice fed the control diet developed a significantly higher hematocrit (HCT) as compared to Jak2-mutant mice fed the obesity diet (p=0.0043). The HCT remained significantly higher in control mice as compared to obese mice over a 46-week period (p<0.0001). Strikingly at 46 weeks, 5/7 mice (71%) fed the control diet showed signs of MPN as compared with 0/6 mice (0%) fed the obesity diet. Furthermore, overall survival was significantly improved in Jak2-mutant obese mice as compared to Jak2-mutant mice fed a control diet (p=0.03). Subsequently, after the death of most control mice, one Jak2-mutant obese mouse developed an elevated HCT, indicating that obesity prolonged the latency to MPN development in Jak2V617F mice. In the chimeric transplant model, gene set enrichment analysis (GSEA) showed significantly increased interferon alpha (NES=2.55, FDR≈0), interferon gamma (NES=2.26, FDR≈0), and interleukin-6 (NES=2.15, FDR=5.46x10-05) signaling in Jak2-mutant LSK cells isolated from obese animals as compared to wild-type LSK cells, a finding not observed in control mice. Single-cell RNAseq experiments further validated these findings, where we identified interferon induced transmembrane protein 1 (IFITM1) as one of the most differentially upregulated genes in Jak2-mutant LT-HSCs from obese mice (adjusted p-value=1.08x10-54) as compared to wild-type LT-HSCs, and unlike Jak2-mutant LT-HSCs from control mice. Conclusions: To our knowledge, our study is the largest to date to evaluate the relationship between obesity and CH and the first to investigate the negative association between JAK2V617F CH and obesity, using functional studies. We developed a novel Fgd5-CreER-Jak2V617F mouse model, which allowed us to study the transition from CH to overt MPN in an unperturbed, non-irradiated bone marrow niche. Transcriptomic profiling points to activation of the interferon signaling pathway in Jak2-mutant HSC as a possible mechanism by which JAK2-mutant LT-HSC may preferentially exhaust during obesity. These findings support testing pegylated interferon as a potential treatment for individuals with JAK2-mutant CH and a clinical protocol is currently under development at our institution. Ongoing functional studies are investigating additional metabolic targets in LT-HSC with the goal of uncovering novel treatment approaches for JAK2-mutant CH.

Although the involvement of glycan structures in diseases has long been recognized, their detailed and high-throughput investigation has only recently been made possible due to technological advancements. For this reason, glycosylation is a generally understudied phenomenon, however it could provide critical information on the pathobiology of many disorders by virtue of its widespread abundance and critical role in protein function. Here, we focus on myeloid malignancies, conditions for which the survival rates are often poor and curative therapeutic options are generally limited. We review the current literature on (1) N-glycosylation of major hematopoietic growth receptors found mutated in myeloid malignancies, (2) chemoresistance through intracellular glycan-related processes, and (3) mechanisms by which altered N-glycosylation contributes to interactions between myeloid blasts and bone marrow stromal cells leading to niche hijacking. For each topic, we describe the related pathobiology and its (potential) clinical implications. The combination of glycoproteomic and genomic information is expected to result in a deeper molecular understanding of the pathobiology of these diseases, which could subsequently be used for improving prognostication and therapeutic strategies.

CONCLUSIONS About 25% of essential thrombocythemia (ET) patients present with extreme thrombocytosis (ExT), defined as having a platelet count ≥1000 x 10^9/L. ExT patients may have an increased bleeding risk, at least in part associated with acquired von Willebrand disease. There are no specific guidelines for managing ExT patients, and many are started on cytoreduction, even if they are younger and otherwise at low thrombotic risk by the revised International Prognostic Score of Thrombosis for ET (R-IPSET). We analyzed the risk of bleeding and thrombosis in ExT vs. non-ExT ET patients to inform treatment decisions. METHODS : We retrospectively analyzed patients who received targeted gene sequencing at Massachusetts General Brigham / Dana-Farber Cancer Institute from 2014-2021 and met WHO 2016 criteria for ET. We performed medical record review and recorded information on treatment, diagnosis, blood counts, and von Willebrand antigen and activity levels. We abstracted the first major bleed, clinically relevant non-major bleed (CRNMB), and thrombotic event in a patient's disease course. Bleeding was categorized using the International Society on Thrombosis & Haemostasis definitions.We determined the cumulative incidence rates of bleeding and thrombosis in ExT and non-ExT patients with death as a competing event and compared these rates using a Gray test. Risk factors for bleeding and thrombosis were evaluated using univariate and multivariate Fine Gray models. We identified 408 patients diagnosed with ET, of which 344 had a platelet count reported at diagnosis. There were 96 ExT patients and 248 non-ExT patients. Median follow-up was 11 and 6 years, and median platelet count at ET diagnosis was 1194 and 651, respectively. ExT patients, compared with non-ExT patients, were more likely to be younger (51 vs. 58 years old, p<.001), have the CALR mutation (49% vs. 23%, p<.001), have lower IPSET-thrombosis risk (65% vs. 49% very low or low risk, p<.001), and have lower von Willebrand activity levels at diagnosis or in disease course (60% vs. 94%, p<.001, Table 1). There were 7 (7%) major bleeds and 7 (7%) CRNMBs in the 96 ExT patients, and 11 (4%) major bleeds and 22 (9%) CRNMBs in the 248 non-ExT patients (p=.29 for major bleeding, p=.83 for CRNMB). 10 year cumulative incidence of major bleed and CRNMB was not different in ExT vs. non-ExT patients (4% vs. 5%, p=.68 for major bleed; 3% vs. 10%, p=.27 for CRNMB, Figure 1). Despite 65% of ExT patients in our sample categorized as very low or low risk for thrombosis, these patients were more likely to be treated with cytoreduction compared to non-ExT patients (63% vs. 50%, p=.21 for very low risk; 81% vs. 51%, p=.022 for low risk). However, there was no difference in the 10 year cumulative incidence of major bleeding and CRNMB in ExT patients who were treated vs. not treated with cytoreduction (6% vs. 0%, p=.13 for major bleed; 2% vs. 8%, p=.13 for CRNMB). There was also no difference in the cumulative incidence of bleeding in ExT and non-ExT patients who were not treated with cytoreduction (untreated or aspirin alone, 0% vs. 4%, p=.34 for major bleed; 8% vs. 9%, p=.75 for CRNMB). In univariate analysis, diabetes mellitus and thrombosis during disease course were significantly associated with increased combined major bleeding and CRNMB. In a multivariable analysis adjusting for these factors in addition to platelet count ≥1000 x 10^9/L and age at ET diagnosis, first von Willebrand factor activity, aspirin and cytoreduction during disease course, and JAK2 mutation status, diabetes mellitus (HR 3.45, 95% CI 1.31-8.72) and thrombosis during disease course (HR 2.92, 95% CI 1.41-6.03) remained as significant risk factors for bleeding. Cytoreduction was not significantly associated with a lower rate of bleeding in multivariable analysis (HR .42, 95% CI .09-3.06). 10 year cumulative incidence of thrombosis was not different in ExT vs. non-ExT patients (22% vs. 26%, p=.94). We found no difference in the cumulative incidence of bleeding and thrombosis in ET patients with ExT and importantly, no clear role for initiation of cytoreductive therapy in these patients who are otherwise low risk by R-IPSET. This confirms the current recommendations to initiate cytoreduction based on thrombotic risk and suggests that ExT alone should not be an indicator for cytoreduction.

EDITORIAL article Front. Mol. Biosci., 13 March 2024Sec. Genome Organization and Dynamics Volume 11 - 2024 | https://doi.org/10.3389/fmolb.2024.1388983

Background: Studies have shown an increased incidence of myeloid neoplasia (MN) in patients with sickle cell disease (SCD). It is hypothesized that increased erythropoietic stress may influence acquisition of molecular drivers of MN in SCD. However, present understanding of MN presentation and outcomes in SCD and other hereditary hemoglobinopathies and hemolytic anemias (HHAs) is limited to case reports. Systematic evaluation of MNs in patients with HHAs will better contextualize the risk of MN in this population relative to non-HHA controls. Methods: Patients with HHAs were identified from the Mass General Brigham Research Patient Data Repository using ICD-10 codes. Evaluated HHAs included sickle cell disease (SCD), sickle cell trait (SCT), thalassemia major, thalassemia trait, other hemoglobinopathy traits, and hereditary spherocytosis or elliptocytosis. A final HHA-MN cohort included patients with laboratory confirmed HHA diagnoses as well as incident myelodysplastic syndrome (MDS), chronic myelomonocytic leukemia (CMML), Ph+ chronic myelogenous leukemia (Ph+ CML), Ph negative myeloproliferative neoplasms (Ph- MPNs) or acute myeloid leukemia (AML). A control cohort of non-HHA MN patients was obtained from the Dana-Farber Hematologic Malignancies Data Repository. Clinical, molecular, and survival data were summarized and compared between HHA-MN and controls. Results: The HHA-MN cohort included 62 patients with HHAs and incident MN (Table 1). HHAs included SCD (n = 6, 9.7%), SCT (n = 9, 14.5%), thalassemia trait (n = 36, 58.1%), thalassemia major (n = 2, 3.2%), hereditary spherocytosis or elliptocytosis (n = 4, 6.5%), and other hemoglobinopathy traits (n = 5, 8.1%). MN was diagnosed at a median age of 61 years [IQR: 48, 70] in the HHA-MN cohort and incident MNs included AML (n = 12, 19.4%), MDS/CMML (n = 22, 35.4%), Ph+ CML (n = 6, 9.7%), and Ph- MPNs (n = 22, 35.4%). A control population of 3388 MN patients without HHA included cases of AML (n = 754, 22.2%), MDS/CMML (n= 1272, 37.5%), Ph+ CML (n = 354, 10.4%), and Ph- MPNs (n = 1008, 29.7%). Compared to HHA patients, control patients were older at MN diagnosis (median age 65 years [IQR: 56, 73], p = 0.0071), driven by significantly younger ages at AML diagnosis for HHA-MN patients (median age 50 years [IQR: 33, 63] vs 65 years [IQR: 57, 73], p =0.0044). AML was the most common MN in SCD (n = 4, 66.7%) whereas incident MDS/CMML was the predominant MN among individuals with thalassemia trait (n = 12, 33.3%) or hereditary spherocytosis and elliptocytosis (n = 2, 50%). Diagnostic cytogenetic data were available for 45 patients in the HHA-MN cohort (72.6% of cohort). Complex karyotype, defined as 3 or more chromosomal aberrations, was observed in 3 of 11 (27.3%) AML, 3 of 19 (15.8%) MDS/CMML and 1 of 17 (5.9%) Ph- MPN cases. Of the 33 HHA-MN patients with available diagnostic next generation sequencing, 60.6% had 2 or more pathogenic mutations at MN diagnosis. Including 2 cases of TP53-mutant AML in patients with SCD, the majority (90%) of AML and 38.9% of MDS/CMML diagnosed in HHA-MN patients were intermediate to adverse risk by ELN criteria or intermediate to very high risk by IPSS-R, respectively. High risk disease was also common in our control MN cohort, where 88.3% of AML cases were intermediate or adverse risk by ELN criteria and 55.9% of MDS/CMML cases were intermediate to very high risk by IPSS-R. Survival varied by subtype of HHA and MN (Figure 1). Survival was lowest for SCD (33.3% at 60-months) and highest for thalassemia traits (67.1% at 60 months). In HHA patients compared to controls, 60-month survival was 65.0% overall (vs 56.4%), 14.3% for AML (vs 23.8%), 51.0% for MDS/CMML (vs 43.9%), 90.5% in Ph- MPN (vs 80.6%), and 100% in Ph+ CML (vs 89.4%). Within the limitations of our sample size, survival differences were not statistically significant. Conclusion: In this retrospective analysis, we observe younger age of diagnosis of myeloid malignancy for patients with HHAs compared to non-HHA controls. Inferior survival in SCD and thalassemia major is largely explained by a higher proportion of AML diagnoses in patients with SCD. AML and MDS/CMML diagnosed in HHA patients tended to be intermediate or high-risk disease, with survival similar to the non-HHA control population that was also enriched for higher risk patients. Validation of our findings in an expanded cohort of HHA patients is currently underway.

MPNs including essential thrombocythemia (ET), polycythemia vera (PV), and myelofibrosis (MF) are characterized by JAK2, CALR and MPL mutations. Additional mutations outside these driver genes also co-occur. To better understand interactions between phenotypic driver mutations and concomitant mutations, we evaluated patterns of somatic mutation acquisition in 1301 MPN (535 ET, 392 PV, 331 MF, and 43 MPN NOS) patients who underwent clinical next generation sequencing (NGS) at our institution. For this analysis, we focused on a single NGS panel obtained closest to MPN diagnosis. Median age at MPN diagnosis was 59 years, with MF patients older than ET/PV patients (p<4e-11). 67%, 17%, and 4.5% of patients had mutations in JAK2, CALR, or MPL. The average variant allele fraction (VAF) for JAK2, CALR and MPL was 48%, 34%, and 40%. 52% of patients had pathogenic mutations in an additional gene (range 0-7), most frequently TET2, DNMT3A, and ASXL1. PV patients were more likely to have a TET2 mutation compared with JAK2-mutated ET patients (42% vs 25%). Mutation combinations that occurred together more frequently than expected (p<1e-3) included DNMTA-TET2, ASXL1-TET2, ASXL1-DNMT3A, and ASXL1-SRSF2. We then investigated how these concomitant mutations interacted with the MPN driver mutation and each other. JAK2-mutatedMPN patients with no concomitant mutations were younger than JAK2-mutated patients with TET2 or ASXL1 mutations. In contrast, age distributions were similar between JAK2-only patients and those with JAK2 and DNMT3A mutations (Fig 1A). This observation held when looking within PV and ET, and whether patients with single or multiple concomitant mutations were considered. Since DNMT3A, TET2 and ASXL1 clonal hematopoiesis is associated with increasing age, the similar age distribution of JAK2-only and JAK2-DNMT3A co-mutant patients was an unexpected finding, and worthy of further study. We evaluated the fraction of JAK2-mutated patients with a given concomitant mutation as a function of age. The fraction of patients with TET2 and ASXL1 mutations increased with age, but not the fraction of patients with DNMT3A mutations (95% CI of slope: 4e-3 to 7e-3 for TET2, 0 to 3e-3 for DNMT3A). Similarly, when we evaluated the fraction of JAK2-mutated patients with a given concomitant mutation as a function of JAK2 VAF, the fraction of patients with TET2 and ASXL1 mutations increased as JAK2 VAF increased, but the fraction of patients with DNMT3A mutations decreased (Fig 1B; 95% CI of slope: 1e-3 to 3e-3 for TET2, -2e-3 to 0 for DNMT3A). One potential explanation for this is that JAK2 and DNMT3A mutations occur as independent clonal acquisitions and that the JAK2-mutant clone “out-competes” the DNMT3A-mutant clone over time. At JAK2 VAFs of 50% and 100%, the ratio of TET2:JAK2 VAFs clustered at 1.0 and 0.5, suggesting frequent presence of a dominant JAK2-TET2 clone in all cells with heterozygous or homozygous JAK2 mutations. ASXL1 mutations occurred mostly in patients with JAK2 VAF > 50%, consistent with it being a subclonal acquisition. In contrast, DNMT3A concomitant mutations were more likely to occur in patients with JAK2 VAF<50% vs >50%, again suggesting that DNMT3A clones diminish as JAK2 clones expand. 6/45 DNMT3A mutations that occurred as the sole concomitant mutation were R882 hotspot mutations. However, DNMT3A was more likely to co-occur at JAK2 VAF>50% in patients with >1 concomitant mutation (most commonly TET2). This suggests that in the context of JAK2-mutant driven MPNs, DNMT3A subclones may require the presence of an additional mutation to expand. In summary, we found DNMT3A mutations behave differently from TET2 and ASXL1 mutations in JAK2-mutant MPNs. Patients with JAK2-mutant MPN are less likely to have DNMT3A mutations as JAK2 VAF increases. Compared with TET2 or ASXL1, DNMT3A appears more likely to exist as an independent clone in JAK2-mutant MPN, with our data suggesting the JAK2-mutant clone exhibits increased fitness compared to the DNMT3A-mutantclone over time. However, DNMT3A mutations persist when accompanied by other concomitant mutations (particularly TET2). These results raise interesting questions regarding clonal competition in MPN. Since all our analyses were performed from single clinical NGS panels, our findings require additional validation, including with multi-gene single-cell genotyping, which we are currently pursuing.

DISCUSSION Myeloproliferative neoplasms (MPNs) including essential thrombocythemia (ET), polycythemia vera (PV), and myelofibrosis (MF) are characterized by increased risk of arterial and venous thrombosis. Cardiovascular risk factors (CV RFs) including hypertension, hyperlipidemia, diabetes, smoking, and obesity likely contribute to thrombotic risk, but the exact incidence of these risk factors and the impact of CV RF modification in MPNs is less clear. The purpose of this study was to determine the prevalence of baseline CV RFs in MPN patients, investigate their association with genomic profiles, and evaluate their effect on long-term outcomes. We retrospectively analyzed patients who received targeted gene sequencing at Massachusetts General Brigham / Dana Farber Cancer Institute (N=977) from 2014-2023, and met WHO 2016 criteria for PV, ET, MF, or pre-fibrotic MF. CV RFs were identified through ICD-9 or 10 codes present prior to MPN diagnosis, and defined as hypertension, hyperlipidemia, diabetes, current smoking status, or BMI>30. Patient and treatment characteristics were described with summary statistics. Genomic profiles were compared between ET, PV, and MF patients with vs without a CV RF. Primary outcome was overall death. Secondary outcomes were venous thromboembolism, arterial thrombosis (including myocardial infarction and stroke), and transformation to MF or acute myeloid leukemia (AML). We calculated cumulative incidence functions of arterial/venous thrombosis as well as overall survival in patients with or without a CV RF. Hazard ratios (HR) were estimated for outcomes using Cox proportional hazards regression. Our cohort contained 399 (39.6%) ET, 312 (31.0%) PV, and 237 (23.5%) MF or pre-fibrotic MF patients. The median age at diagnosis was 58.5 years, and 47.9% of patients were male. The overall prevalence of hyperlipidemia, hypertension, and diabetes at MPN diagnosis was 16%, 20%, and 8%. The average BMI at diagnosis in all MPNs was 27.28, with 64% and 23% of patients having a BMI of >25 and >30. Six percent of MPN patients were current smokers at time of diagnosis, compared to 39% former smokers and 56% never smokers. ET and PV patients with ≥1 CV RF at MPN (N = 234, 32.9%) diagnosis were older (mean age 61.1 vs 52.1 years, p<0.001), and more likely to be male (50.9% vs 41.3%, p=0.02), non-White (12.4% vs 7.6%, p=0.015), and have a prior history of atherosclerotic disease (16.2% vs 4.4%, p<0.001), thrombosis (13.7% vs 5.0%, p<0.001), and heart failure (2.1% vs 0.4%, p=0.042). MF patients with ≥1 CV RF (N = 120, 50.6%) were also more likely to be older (mean age 67.5 vs 60.4 years, p<0.001), male (66.7% vs 53.9%, p=0.047), and have a prior history of atherosclerotic disease (19.2% vs 6.0%, p=0.003) (Table). Results of targeted gene sequencing closest to MPN diagnosis were analyzed. A similar proportion of driver ( JAK2, CALR, MPL) and concomitant ( TET2, ASXL1, DNMT3A, SRSF2, SF3B1, U2AF1, ZRSR2) mutations were seen in MPN patients with or without a CV RF. However, patients with ET or PV with ≥1 CV RF had a lower variant allele fraction (VAF) of their driver mutation (mean 41.0% vs 48.1%, p = 0.004), which was primarily driven by JAK2 (mean 43.5% vs 52.0%, p=0.003). In ET and PV, after adjusting for variables significant on univariate analysis, the presence of ≥1 CV RF was associated with higher risk of death from any cause (HR 1.73, 95% CI 1.08-2.76) and arterial thrombosis (HR 2.33, 95% CI 1.22-4.42). Among patients with MF, the presence of ≥1 CV RF was not associated with increased risk of death when adjusted for age, sex, and prior thrombosis (HR 1.36, 95% CI 0.84 - 2.20). MPN patients with ≥1 CV RF did not have increased rates of MF or leukemia progression (Fig). In our study, CV RFs are common among patients with MPN. CV RFs were associated with adverse outcomes, including death and thrombosis among patients with ET or PV. We found no differences in the molecular profiles in MPN patients with or without CV RFs, although a lower JAK2 VAF was seen in patients without CV RFs, which will need to be explored further. MPN patients with a CV RF had significantly worse overall survival and cumulative arterial thrombosis rates, although the presence of CV RFs does not impact MF or leukemia progression. However, our results highlight the importance of addressing CV RFs in MPN care to improve morbidity and mortality.

Introduction: TP53 mutations in patients with myeloproliferative neoplasms (MPN) are associated with poor prognosis, including progression to blast phase MPN. However, low variant allele fraction (VAF) TP53 mutations have been reported to remain stable over years in chronic phase MPN. A major unmet clinical need in MPN is the ability to discriminate patients with TP53-mutant MPN who are at high-risk of secondary AML (sAML) and warrant immediate intervention from those who are at lower risk of sAML in whom active surveillance can be employed. Therefore, we sought to identify parameters associated with leukemic transformation and overall survival in the context of MPN with genetic aberrations in TP53. Materials and Methods: We retrospectively analyzed a cohort of 947 MPN patients from the Dana-Farber Cancer Institute Hematologic Malignancies Data Repository (HMDR) with at least one clinical next-generation sequencing (NGS) panel performed. Patient characteristics such as age at MPN diagnosis, gender, MPN subtype and driver mutations were recorded. Furthermore, information about the course of disease was extracted including occurrence of sAML and overall survival (Figure 1). We also analyzed type and number of additional mutations as well as cytogenetics. With respect to TP53-specific parameters, we evaluated the number of TP53 mutations, TP53 VAF, loss of heterozygosity (LOH) at the TP53 locus, phenotypic annotations of TP53 (i.e. PHANTM score) and 17p deletion. We defined “multi-hit” TP53 as the presence of two or more TP53 mutations, TP53 VAF higher than 50%, TP53 mutation plus 17p deletionor TP53 mutation and documented LOH. Results: A total of 947 patients were analyzed, of which 40 harbored at least one detectable TP53 mutation. A total of 13 patients were found to have a multi-hit TP53 mutations defined by > 50% VAF in 6 patients, two or more TP53 mutations in 5 patients and TP53 mutation + 17p deletion in 5 patients. The MPN diagnosis at time of TP53 mutation detection was post ET/PV myelofibrosis (secondary MF) (n=23, 58%), primary myelofibrosis (MF) (n=7, 18%), pre-fibrotic MF (n=2, 5%), essential thrombocythemia (ET) (n=6, 15%) and polycythemia vera (PV) (n=2, 5%). Two patients with ET and one patient with PV did not have a concurrent in-house bone marrow biopsy performed at the time the TP53 mutation was detected. Two patients with ET developed sAML within 12 months of TP53 mutation detection, without prior mention of fibrosis. Age at first MPN diagnosis was not significantly different between patients with or without TP53 mutation. The average time from initial MPN diagnosis to detection of the first TP53 mutation was 9 years (range: 0-33 years). The most common MPN driver mutation among patients with TP53 mutations was JAK2 (75%), followed by CALR (13%)and MPL (5%). Out of all TP53-mutated patients, 8% showed a triple negative status. The most frequent additional mutations among patients with TP53 mutations were TET2 (25%), U2AF1 (15%), ASXL1 (13%), and DNMT3A (10%). There was no significant difference between single-hit and multi-hit TP53 status regarding MPN subtype, driver mutations and co-mutations (Table 1). Seven patients (single-hit: 15%, multi-hit: 23%) with a TP53 mutation developed sAML during the course of their disease, compared with only 3% of all patients without a TP53 mutation and 50% (single-hit: 41%, multi-hit: 69%) were deceased at the time of the last follow-up compared to 18% of all patients without a TP53 mutation. We focused on overall survival from the initial MPN diagnosis and considered whether patients developed bone marrow fibrosis during their disease course (Figure 1). Survival did not differ significantly between single-hit TP53 and patients with multi-hit TP53 (p=0.2), but survival did differ significantly between multi-hit TP53 patients and TP53 wildtype patients with MF/prefibrotic MF/Secondary MF (p=0.02) as well as compared to all MPN patients without a TP53 mutation (p<0.001). Survival was not significantly different between single-hit TP53 and TP53 wildtype MF/prefibrotic MF/Secondary MF patients (p=0.4). Conclusions: In a large cohort of 947 molecularly characterized MPN patients, 4% of the cohort developed a TP53 mutation during their course of disease. 18% of all TP53-mutant patients developed sAML with an adverse effect on overall survival for patients with multi-hit but not single-hit TP53 mutations.

Novel therapeutics are urgently needed to prevent opportunistic infections in immunocompromised individuals undergoing cancer treatments or other immune-suppressive therapies. Trained immunity is a promising strategy to reduce this burden of disease. We previously demonstrated that mesenchymal stromal cells (MSCs) preconditioned with a class A CpG oligodeoxynucleotide (CpG-ODN), a Toll-like receptor 9 (TLR9) agonist, can augment emergency granulopoiesis in a murine model of neutropenic sepsis. Here, we used a chimeric mouse model to demonstrate that MSCs secrete paracrine factors that act on lineage-negative c-kit+ hematopoietic stem cells (HSCs), leaving them "poised" to enhance emergency granulopoiesis months after transplantation. Chimeric mice developed from HSCs exposed to conditioned media from MSCs and CpG-ODN-preconditioned MSCs showed significantly higher bacterial clearance and increased neutrophil granulopoiesis following lung infection than control mice. By Cleavage Under Targets and Release Using Nuclease (CUT&RUN) chromatin sequencing, we identified that MSC-conditioned media leaves H3K4me3 histone marks in HSCs at genes involved in myelopoiesis and in signaling persistence by the mTOR pathway. Both soluble factors and extracellular vesicles from MSCs mediated these effects on HSCs and proteomic analysis by mass spectrometry revealed soluble calreticulin as a potential mediator. In summary, this study demonstrates that trained immunity can be mediated by paracrine factors from MSCs to induce neutrophil-trained immunity by reprogramming HSCs for long-lasting functional changes in neutrophil-mediated antimicrobial immunity.

RNA splicing factor gene mutations are recurrently found in BCR::ABL1 negative myeloproliferative neoplasms (MPN). To understand the interactions between spliceosome gene mutations and MPN phenotypic driver mutations ( JAK2, CALR and MPL), we investigated their patterns of co-occurrence or mutual exclusivity in a cohort of 990 MPN patients with comprehensive molecular profiling and clinical information from the Hematologic Malignancies Data Repository (HMDR). As expected, MPN driver mutations ( JAK2, CALR and MPL) were mutually exclusive (N=658, N=168, N=53 respectively, Log2 OR≤-3.5 and Padj. ≤0.005 using pairwise Fisher's exact tests with Benjamin Hochberg multiple testing correction). Additionally, cases with co-occurring mutations in two RNA splicing factors, SF3B1, U2AF1 and SRSF2 (N=44, N=49, N=44) respectively were uncommon (2/135 splicing factor mutated cases) . We further found that U2AF1 and SRSF2 mutations co-occurred with JAK2 (log2 OR=1.7, Padj.=0.009) and MPL (log2 OR=2.1, Padj.=0.006) respectively, but rarely with CALR mutations (N=2 CALR-U2AF1 log2 OR=-2.3 Padj.=0.025, N=2 CALR-SRSF2 log2 OR=-2.2 Padj.=0.048). In contrast, CALR-SF3B1 mutations occurred at the expected frequency (Padj.=0.11). To explore the molecular settings where these mutations co-occur, we investigated the variant allele frequencies (VAFs) of CALR, U2AF1, SRSF2 and concomitant mutations in the CALR-U2AF1 (N=2) and CALR-SRSF2 (N=2) patients. The VAFs for the mutations detected in the first patient with CALR-U2AF1 co-mutation were: CALR 0.26, U2AF1 0.45, and TET2 0.41. Since CALR and splicing factor mutations are typically heterozygous, this suggests the CALR mutation was sub-clonal to the U2AF1 mutation and that U2AF1 and TET2 mutations were likely present in the same cell. In the second CALR-U2AF1 co-mutant case, VAFs were: CALR 0.31, U2AF1 0.27, and TP53 0.16. It is possible that in this case, the CALR and U2AF1 mutations may have occurred in independent clones or may have occurred in the same cell in the absence of other pathogenic mutations. For the two CALR-SRSF2 patients, data from three samples revealed VAFs of 0.23-0.46 for CALR and 0.21-0.58 for SRSF2 mutations. Considering their typical heterozygous occurrence, we inferred those individual cells harbored both CALR and SRSF2 mutations. Notably, in samples with high VAFs (>0.37) for both CALR and SRSF2 in two separate patients, additional high VAF pathogenic mutations were observed: 0.35 for ASXL1 and 0.46 for RUNX1 respectively. As ASXL1 and RUNX1 mutations also typically occur in a heterozygous manner, we hypothesized that when CALR and SRSF2 mutations are present in the same cell, another pathogenic co-mutation may be required for survival of the clone. To test our hypothesis, we FACS-sorted single lineage negative CD34 positive bone marrow cells from the first CALR- SRSF2 co-mutant case into 96 wells plates containing methylcellulose and hematopoietic growth factors (CFU-GEMM). DNA was harvested from single-cell colonies and genotyped for the following mutations using polymerase chain reaction (PCR) followed by gel electrophoresis and Sanger sequencing: CALR c.1154_1155insTTGTC p.K385fs*, SRSF2 c.284G>T p.P95H, ASXL1 c.1926_1927insG p.G642fs*, EZH2 c.2188_2212delAAAAAACAGCTCTTCGCCAGTCTGG p.F729fs*. Bulk NGS VAFs were 0.46, 0.38, 0.35, and 0.17, respectively. Out of 73 colonies analyzed, 71 colonies had the CALR insertion 5 mutation detected (see figure). Among these 71 CALR-mutated colonies, SRSF2 was found to co-occur in 35 colonies. Interestingly, all 35 colonies were accompanied by an ASXL1 mutation, and 29 of them had an additional EZH2 mutation. These findings support our hypothesis that CALR and SRSF2 mutations may co-occur in the same cell in the presence of another pathogenic mutation. Our findings offer molecular insights into the context-dependent behavior of splicing factor mutations in MPN. While they can co-occur with JAK2 and MPL mutations, U2AF1 and SRSF2 mutations are largely mutually exclusive with CALR mutations. In addition, our findings generate two main biological hypotheses: (i) a synthetic lethal relationship may exist between CALR mutations and SRSF2 or U2AF1 mutations, and (ii) this synthetic lethality may be overcome by the presence of additional pathogenic mutation(s) in the cell. Validation studies are ongoing to address these hypotheses. AM and AEM contributed equally.

Cancer is driven by somatic mutations that provide a fitness advantage. While targeted therapies often focus on the mutated gene or its direct downstream effectors, imbalances brought on by cell-state alterations may also confer unique vulnerabilities. In myeloproliferative neoplasms (MPN), somatic mutations in the calreticulin (CALR) gene are disease-initiating through aberrant binding of mutant CALR to the thrombopoietin receptor MPL and ligand-independent activation of JAK-STAT signaling. Despite these mechanistic insights into the pathogenesis of CALR-mutant MPN, there are currently no mutant CALR-selective therapies available. Here, we identified differential upregulation of unfolded proteins, the proteasome and the ER stress response in CALR-mutant hematopoietic stem cells (HSCs) and megakaryocyte progenitors. We further found that combined pharmacological inhibition of the proteasome and IRE1-XBP1 axis of the ER stress response preferentially targets Calr-mutated HSCs and megakaryocytic-lineage cells over wild-type cells in vivo, resulting in an amelioration of the MPN phenotype. In serial transplantation assays following combined proteasome/IRE1 inhibition for six weeks, we did not find preferential depletion of Calr-mutant long-term HSCs. Together, these findings leverage altered proteostasis in Calr-mutant MPN to identify combinatorial dependencies that may be targeted for therapeutic benefit and suggest that eradicating disease-propagating Calr-mutant LT-HSCs may require more sustained treatment.

Calreticulin (CALR) mutations are frequent, disease-initiating events in myeloproliferative neoplasms (MPNs). Although the biological mechanism by which CALR mutations cause MPNs has been elucidated, there currently are no clonally selective therapies for CALR-mutant MPNs. To identify unique genetic dependencies in CALR-mutant MPNs, we performed a whole-genome clustered regularly interspaced short palindromic repeats (CRISPR) knockout depletion screen in mutant CALR-transformed hematopoietic cells. We found that genes in the N-glycosylation pathway (among others) were differentially depleted in mutant CALR-transformed cells as compared with control cells. Using a focused pharmacological in vitro screen targeting unique vulnerabilities uncovered in the CRISPR screen, we found that chemical inhibition of N-glycosylation impaired the growth of mutant CALR-transformed cells, through a reduction in MPL cell surface expression. We treated Calr-mutant knockin mice with the N-glycosylation inhibitor 2-deoxy-glucose (2-DG) and found a preferential sensitivity of Calr-mutant cells to 2-DG as compared with wild-type cells and normalization of key MPNs disease features. To validate our findings in primary human cells, we performed megakaryocyte colony-forming unit (CFU-MK) assays. We found that N-glycosylation inhibition significantly reduced CFU-MK formation in patient-derived CALR- mutant bone marrow as compared with bone marrow derived from healthy donors. In aggregate, our findings advance the development of clonally selective treatments for CALR-mutant MPNs.

Polycythemia vera (PV) is characterized by marked erythrocytosis, most commonly managed with therapeutic phlebotomy or cytoreduction. Conversely, anemia is often a feature of myelofibrosis (MF), where it is associated with poor prognosis. The transition from PV to MF is associated with the acquisition of additional somatic mutations in several gene classes, including RNA splicing factors. Since mutations in the RNA splicing factor SRSF2 are associated with anemia in myelodysplastic syndrome (MDS), we set out to investigate how SRSF2 mutations impact erythropoiesis in multiple contexts, including at steady-state, stress-induced, and in the setting of MPN. We first studied the consequences of Srsf2P95H mutations on steady-state erythropoiesis. Primary Mx-Cre Srsf2P95H/+ mice had a reduced red blood cell count (RBC, P<0.001), increased mean corpuscular volume (MCV, P<0.001), and reduced percentage of early erythroid progenitors (pre-CFU-E) in the bone marrow (BM, P=0.04) compared to Mx-Cre mice, consistent with published data (Kim et al., 2015). Moreover, Srsf2P95H/+ mice had a reduced percentage of proliferating pre-CFU-E cells (P=0.03). We next studied the impact of Srsf2P95H mutations on stress-induced erythropoiesis, using a phenylhydrazine (PHZ)-induced hemolysis model. We treated primary Mx-Cre and Mx-Cre Srsf2P95H/+ mice with 30 mg/kg PHZ on days 0 and 1. As expected, RBCs were reduced by ~40% confirming effective hemolysis on day 3. On day 6, PHZ-treated control mice had higher reticulocyte counts compared to Srsf2P95H/+ mice (P=0.005), while the inverse was true on day 9 (P=0.03). Consistently, the frequency of BM pre-CFU-E cells was increased in Srsf2P95H/+ mice vs controls on day 9 (P=0.04). In aggregate, these data indicate that Srsf2P95H/+ mice have a slower recovery to PHZ-induced anemia. Moreover, late erythroid progenitors (CFU-E) were reduced in PHZ-treated Mx-Cre Srsf2P95H/+ vs control mice on day 4 post PHZ (P=0.04). Also, the percentage of BM erythroblasts was 24% reduced (P=0.04), mostly due to a reduction in stage III erythroblasts, and spleen weights were 19% lower in Srsf2P95H/+ vs controls on day 4 (P=0.01). These data suggest defects at the stage III erythroblast, and erythroid progenitor levels due to Srsf2P95H. Our next goal was to understand how Srsf2 mutations affect Jak2-mutant MPN. In line with the data outlined above, primary Mx-Cre Jak2V617F/+Srsf2P95H/+ mice had lower RBC (Fig. A, P=0.002) and higher MCV values (P<0.001) compared to Mx-Cre Jak2V617F/+ mice. These data were recapitulated in a BM transplant model confirming that impaired erythropoiesis is cell intrinsic. Moreover, pre-CFU-E and CFU-E levels were lower in the BM and spleen of Jak2V617F/+Srsf2P95H/+ vs Jak2V617F/+ transplant mice ~30 weeks post pIpC (BM P=0.003 and 0.01, spleen 0.02 and 0.005, respectively). Compared to wild-type mice, mice transplanted with Jak2V617F/+ cells had an elevated fraction of stage II and reduced fraction of stage III erythroblasts, which was partially normalized in Jak2V617F/+Srsf2P95H/+ mice. Finally, spleen weights were >50% lower in Jak2V617F/+Srsf2P95H/+ vs Jak2V617F/+ transplant mice (P<0.001). These data show that Srsf2P95H diminishes Jak2V617F-driven erythropoiesis at the progenitor level and at erythroblast stages II and III. Finally, we interrogated a clinical dataset of 301 JAK2-mutant MPN patients (PV n=157; ET n=129; pre-fibrotic MF n=15) with comprehensive molecular profiling and laboratory results. Eight patients carried both, an SRSF2 mutation (VAF>5%) and a JAK2 mutation. JAK2-SRSF2 co-mutant patients had significantly reduced hemoglobin values compared to JAK2-mutant patients (Fig. B, P=0.01), in agreement with our mouse models. Together, our study shows that Srsf2P95H impairs erythropoiesis in the BM and spleen at multiple differentiation levels. These data contribute to a better understanding of phenotypic differences between JAK2-mutant MPN patients. Since several genes implicated in erythropoiesis and heme metabolism have been shown to be misspliced in SRSF2-mutant MDS, we are currently analyzing RNA-sequencing data from wild-type and Srsf2P95H/+ CFU-E cells, with the expectation that these data will provide insights into the underlying mechanisms driving anemia. These data may ultimately help identify novel therapeutic targets to alleviate anemia in SRSF2-mutant MPN and MDS, which is an area of unmet clinical need. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Background: Mutations in the endoplasmic reticulum (ER) chaperone calreticulin (CALR) are frequent and disease-initiating in myeloproliferative neoplasms (MPN). These mutant CALR proteins have impaired chaperone function. Concordant with this, transcriptional upregulation of the unfolded protein response (UPR) has been reported in patients with CALR-mutated MPN. However, it is not understood how CALR-mutated cells counter-balance ER stress resulting from impaired chaperone function. Despite the frequency of CALR mutations in MPN, there are currently no treatment strategies to preferentially target CALR-mutant cells over healthy cells. Aims: To determine the mechanisms by which mutant CALR cells alleviate ER stress and to exploit these mechanisms selectively by pharmacological intervention in vivo. Methods: Long-term HSCs (LT-HSCs) from CalrΔ52 knockin mice were isolated for RNA-Seq. CalrΔ52 bone marrow (BM) was differentiated ex vivo, subsequently enriched for megakaryocytes, and subjected to quantitative proteomic analysis. Pathway analyses of CALR-mutant megakaryocyte progenitors (MkPs) were performed on Genotyping of Transcriptome (GoT) data from patients with Essential Thrombocytosis (ET). CalrΔ52 VAVCre mice and chimeric transplanted CalrΔ52 mice were treated with a proteasome inhibitor, an IRE1a inhibitor, or both. Results: To identify dysregulated pathways in CalrΔ52/Δ52 LT-HSCs, we performed RNA-seq and found that Xbp1s downstream of IRE1a and the proteasome pathway was significantly upregulated in CalrΔ52/Δ52 animals. We confirmed elevated XBP1s protein level in heterozygous CalrΔ52 knockin mice by flow cytometry. Since elevated and abnormal megakaryopoiesis are hallmarks of CALR-mutant MPN, we performed quantitative proteomics on megakaryocyte-enriched cells from CalrΔ52 mice. We confirmed upregulation of both proteasome and ER chaperone proteins in Calr-mutant cells. To verify these findings in human MPN, we interrogated a published GoT data set and found that in addition to XBP1s, the proteasome pathway was also differentially upregulated in CALR-mutated patient MkPs compared to the wildtype MkPs (p<2e-16). To investigate the functional consequences of IRE1a or proteasome inhibition, we treated CALR-mutant hematopoietic cell lines with KIRA6 (IRE1a inhibitor) and/or bortezomib (proteasome inhibitor) and found CALR-mutant cells to be differentially sensitive as compared to isogenic controls. We subsequently found that treatment with bortezomib leads to the accumulation of misfolded and ubiquitinylated proteins in mutant Calr BM, resulting in pro-apoptotic priming. In vivo treatment of CalrΔ52/+ mice with bortezomib resulted in a significant reduction in platelets. Finally, we combined IRE1a and proteasome inhibition in a chimeric BM transplantation in vivo model using CD45.2 CalrΔ52/+ MxCre GFP and CD45.1 wild-type competitor cells. Following combination therapy, we found a significant reduction in Calr-mutant donor chimerism, specifically in LT-HSCs and platelets, accompanied by reductions in LT-HSC frequency and platelet count. Summary/Conclusion: In summary, we have found that disrupted proteasis in CALR-mutated MPN cells results in a dependency on the IRE1-XPB1 axis of the UPR and proteasome activity. Using pre-clinical MPN models, we found that combined inhibition of both pathways in vivo normalizes important features of MPN and preferentially targets CalrΔ52 over wild-type cells. These findings highlight combined inhibition of IRE1 and the proteasome as a promising therapeutic strategy in CALR-mutant MPN.