The major challenges of gene therapy for hemophilia A using adeno-associated virus (AAV) vectors are reducing vector doses and the long-term maintenance of stable factor VIII (FVIII). In this study, we developed engineered human B-domain-deleted FVIIIs (FVIIISQ) with enhanced secretion and coagulation potential. Intracellular accumulation was markedly reduced in some engineered FVIIISQ, resulting in reduced unfolded protein responses. The administration of AAV vectors carrying engineered FVIIISQ to hemophilia A mice resulted in ∼8-fold higher FVIII activity and 4-fold higher FVIII antigen levels compared with wild-type FVIIISQ administration. The specific FVIII activity of the engineered FVIIISQ was 3.6 times higher than that of the wild-type FVIIISQ, and its binding to activated coagulation factor IX was significantly enhanced, which is supported by the structural analysis. In macaques, the administration of AAV5 vector carrying the engineered FVIIISQ without CpG sequences resulted in a supraphysiological increase in plasma FVIII activity at a dose one-thirtieth that of valoctocogene roxaparvovec (2 × 1012 vector genome per kg). The engineered FVIIISQ may thus provide stable, long-term therapeutic efficacy in AAV-mediated hemophilia A gene therapy even at low doses.

Self-renewal and differentiation are at the basis of hematopoiesis. Although it is known that tight regulation of translation is vital for hematopoietic stem cells' (HSC) biology, the mechanisms underlying translation regulation across the hematopoietic system remain obscure. Here, we reveal a novel mechanism of translation regulation in the hematopoietic hierarchy, which is mediated by rRNA methylation dynamics. Using ultralow-input ribosome profiling, we characterized cell-type-specific translation capacity during erythroid differentiation. We found that translation efficiency (TE) changes progressively with differentiation and can distinguish between discrete cell populations, as well as define differentiation trajectories. To reveal the underlying mechanism, we performed comprehensive mapping of the most abundant rRNA modification, 2'-O-methyl (2'OMe). We found that, such as TE, 2'OMe dynamics followed a distinct trajectory during erythroid differentiation. Genetic perturbation of individual 2'OMe sites demonstrated their distinct roles in modulating proliferation and differentiation. By combining CRISPR screening, molecular, and functional analyses, we identified a specific methylation site, 28S-Gm4588, which is progressively lost during differentiation, as a key regulator of HSC self-renewal. We showed that low methylation at this site led to translational skewing, mediated mainly by codon frequency, which promoted differentiation. Functionally, HSC with diminished 28S-Gm4588 methylation exhibited impaired self-renewal capacity ex vivo, and loss of fitness in vivo in bone marrow transplants. Extending our findings beyond the hematopoietic system, we also found distinct dynamics of 2'OMe profiles during differentiation of non-HSC. Our findings reveal rRNA methylation dynamics as a general mechanism for cell-type-specific translation, required for cell function and differentiation.

Therapy resistance in acute myeloid leukemia (AML) remains a major clinical obstacle, particularly because of the persistence of leukemia stem cells (LSC) capable of metabolic adaptation. Although venetoclax (Ven) inhibits oxidative phosphorylation (OXPHOS), we found that Ven-resistant LSC undergo glycolytic reprogramming to bypass OXPHOS inhibition. This metabolic shift is supported by enhanced ribosome biogenesis, which is sustained by upregulated de novo guanine nucleotide biosynthesis. Abundant guanine nucleotides suppress the impaired ribosome biogenesis checkpoint (IRBC), leading to TP53 destabilization and persistent MYC expression. The inhibition of inosine monophosphate dehydrogenases (IMPDH1/2) depletes guanine nucleotides, activates IRBC, stabilizes TP53, represses MYC, and impairs the metabolic shift to glycolysis. This metabolic rewiring disrupts LSC stemness and suppresses the reconstitution of human AML cells in xenotransplantation experiments. Notably, the suppression of LSC stemness was observed regardless of Ven resistance or the TP53 mutational status of AML cells. These findings reveal that mutation-independent TP53 inactivation is involved in resistant AML and suggest that targeting guanine nucleotide biosynthesis may offer a clinically actionable strategy to eradicate therapy-resistant LSC.

The transcription factor BCL11A is a genetically and clinically validated regulator of the fetal-to-adult hemoglobin switch in human erythroid cells. CRISPR editing of an intronic enhancer within the BCL11A gene reactivates fetal hemoglobin (HbF) in adult erythroid cells, serving as the first CRISPR-based therapy for β-hemoglobinopathies. However, the molecular basis for the remarkable efficacy of CRISPR-mediated enhancer ablation remains elusive. Here, we describe a new genome architecture, an enhancer-dependent chromatin rosette, that is essential for epigenetic insulation and the developmentally regulated, hematopoietic lineage-specific expression of BCL11A. CRISPR-mediated disruption of the BCL11A erythroid enhancer impairs transcription of enhancer-driven RNAs and NIPBL-dependent cohesin loading, leading to the destabilization of the rosette structure, loss of chromatin insulation, and epigenetic silencing of BCL11A. Moreover, targeted depletion of enhancer RNAs using antisense oligonucleotides silences BCL11A by disrupting epigenetic insulation, causing HbF reactivation in adult erythroid cells. These findings uncover an essential role for enhancer-driven epigenetic insulation in transcriptional control, presenting a new strategy for the therapeutic targeting of BCL11A.

Laboratory reference intervals must reflect population diversity for accurate medical decisions. The Duffy-null variant lowers absolute neutrophil counts (ANC), but existing dedicated reference intervals are based on a single African American cohort. The impact across other ethnic groups and regions remains unclear, and no white blood cell count (WBC) intervals exist for Duffy-null individuals. This study aimed to establish and compare Duffy-null ANC and WBC reference intervals across 4 continents. A cross-sectional study was conducted assessing healthy Duffy-null individuals from dedicated cohorts (blood donors in Namibia, Saudi Arabia, and the United Kingdom; primary care patients in the United States) and biobanks (participants from the United Kingdom and the United States). Among 8018 participants (880 from dedicated cohorts and 7138 from biobanks), novel ANC and WBC reference intervals were established (Namibia [ANC, 820/μL to 6370/μL; WBC, 2.51 × 109/L to 9.85 × 109/L]; Saudi Arabia [ANC, 1140/μL to 5290/μL; WBC, 3.72 × 109/L to 10.71 × 109/L]; United Kingdom (ANC, 1185/μL to 5462/μL; WBC, 3.1 × 109/L to 8.8 × 109/L]; the United States [ANC, 1210/μL to 5390/μL; WBC, 3.00 × 109/L to 9.66 × 109/L]), with no significant differences between cohorts. Institutional reference intervals misclassified 27.9% (Namibia), 50.9% (Saudi Arabia), 26.0% (United Kingdom), and 21.7% (the United States) as neutropenic. Biobank analyses confirmed no significant difference in ANC between Black and non-Black Duffy-null participants. Duffy-null individuals consistently exhibit lower ANC and WBC across ethnic groups and regions. Current reference intervals overlook this variation, risking misdiagnosis and health inequities. Implementing Duffy-specific reference intervals is essential for equitable and accurate clinical decisions worldwide.

Acquired resistance to targeted, nonintensive therapies is common in myeloid malignancies. However, the kinetics of selection, the hematopoietic cell compartments in which selection occurs, and the molecular mechanisms underlying selection remain open questions. To address this, we studied the kinetics of clonal and transcriptional responses to combinational therapy with ivosidenib plus venetoclax, with or without azacitidine, across hematopoiesis in 8 patients with IDH1-mutant myeloid malignancy. All 8 patients initially responded to treatment, but 6 relapsed, whereas 2 remained in sustained remission for >4 years. We performed combined high-sensitivity single-cell genotyping and scRNA sequencing in index-sorted sequential patient samples. In all patients, clonal selection occurred rapidly, within 1 to 3 treatment cycles. Clonal selection preceded treatment failure by months to years. Relapse was associated with expansion of either clones harboring newly detected myeloid driver mutations or preexisting minor clones that underwent differentiation delay upon treatment exposure. In both cases, clonal selection occurred within immature cell populations previously shown to contain leukemic stem cell potential. Different genetic alterations within relapse-associated clones converged onto common upregulated transcriptional programs of stemness, branched-chain amino acid catabolism, and genes sensitive to menin inhibition. Importantly, this relapse-associated transcriptional signature was selected within 3 cycles of therapy. In contrast, in both patients remaining in remission, leukemic clones were rapidly eradicated, and replaced by clonal and wild-type hematopoiesis. Overall, in patients treated with ivosidenib combination therapy, rapid clonal selection occurs within the first treatment cycles. In those patients destined to relapse, genetically heterogeneous resistant clones are characterized by common transcriptional programs.

We previously reported a chemogenomics screen that unexpectedly identified phosphatidylinositol-3-phosphate 5-kinase (PIKfyve) as a vulnerable target in multiple myeloma (MM). PIKfyve is an essential regulator of lysosomal function and autophagy. Given the high basal requirement for autophagy in MM for sustainable immunoglobulin synthesis, targeting autophagy holds clinical potential as a novel therapeutic avenue. Here, we report the development and characterization of PIK001 and analogs, potent and selective novel small-molecule inhibitors of PIKfyve. PIK001 demonstrated potent anti-MM activity in vitro, as well as synergistic activity with established anti-MM agents (including venetoclax and selinexor), while retaining efficacy in lenalidomide-resistant models. Multiomic characterization of isogenic cell lines sensitive and resistant to PIK001 identified a catalytic domain mutation (PIKFYVE N1939K) and heterogenous alterations in autophagy capabilities. Importantly, we noted that PIK001 exposure also resulted in significantly increased cholesterol metabolism and upregulation of major histocompatibility complex (MHC) class I expression, with potential implications in tumor immunity. Beyond MM, PIKfyve inhibition also shows selective cytotoxicity in acute myeloid leukemia, melanoma, and renal cancer, highlighting broader therapeutic potential. These findings establish PIKfyve inhibition as a valid target for MM and other hematologic malignancies, provide insights into mechanisms of sensitivity and resistance, and lay the foundation for further preclinical (particularly the role of cholesterol metabolism and tumor immunity) and clinical development.

Aside from allogeneic transplantation, the current standard-of-care approach for higher-risk myelodysplastic syndromes/neoplasms (HR-MDS) remains monotherapy with a hypomethylating agent (HMA), including azacitidine, decitabine, or oral decitabine/cedazuridine. Many attempts using HMA-based combinations have failed to improve upon HMA monotherapy. Although promising efficacy was observed in early-phase clinical trials with several agents, subsequent randomized phase 3 trials failed to confirm improvements in complete response rates or overall survival. In this review, we discuss lessons learned from the recently reported negative trials of azacitidine in combination with eprenetapopt (APR-246), magrolimab, pevonedistat, sabatolimab, tamibarotene, and venetoclax. First, we make a case for emphasizing biological classification rather than disease risk status alone to select patients for HR-MDS trials. Second, we argue that patients with TP53-inactivated MDS and chronic myelomonocytic leukemia should be treated in dedicated clinical trials. Alternatively, if TP53-inactivated MDS is included in HR-MDS trials, then randomization stratification by TP53 inactivation status should be considered. Third, we caution against ignoring signals of excessive toxicity and premature discontinuation of investigational agent observed in early-phase trials. Fourth, we show that the International Working Group (IWG) 2006 response criteria, long used in HR-MDS trials, can both overestimate and underestimate the true therapeutic benefit. Instead, we advocate for using the IWG 2023 response criteria to better capture clinically meaningful benefits in HR-MDS. Lastly, we emphasize the need for the scientific community to access patient-level data and samples from failed phase 3 trials in an efficient and expedited fashion to inform the development of subsequent trials.

Classic Hodgkin lymphoma (cHL) is highly curable with risk-adapted first-line treatment. Due to exceptional efficacy, antiprogrammed cell death protein 1 antibodies (aPD1) are increasingly incorporated into first-line treatment. However, the short- and long-term immune-related adverse event burden in this setting is insufficiently understood. Here, we review the currently available evidence on the feasibility and safety of aPD1 first-line cHL treatment. A more harmonized and complete reporting is critical to enable a detailed understanding and comprehensive assessment of aPD1-related morbidity.

High-grade B-cell lymphoma with 11q aberration (HGBCL-11q) is a rare pediatric non-Hodgkin lymphoma. This study assessed outcome in 90 children with HGBCL-11q. With survival rates ≥95%, patients with HGBCL-11q and no predisposition are candidates for deescalated therapy in future prospective trials.

Acute myeloid leukemia (AML) with TP53 mutations is almost universally refractory to chemotherapy, molecular-targeted therapies, and hematopoietic stem cell transplantation, leading to dismal clinical outcomes. The lack of effective treatments underscores the urgent need for novel therapeutic strategies. Using genome-wide CRISPR/Cas9 dropout screens in isogenic Trp53-wild-type (WT) and Trp53-knockout mouse AML models, combined with transcriptomic and proteomic analyses of AML samples from mice and humans, we identify the XPO7-NPAT (exportin 7-nuclear protein, coactivator of histone transcription) pathway as essential for TP53-mutated AML cell survival. In TP53-WT AML, XPO7 functions as a tumor suppressor by regulating the nuclear abundance of p53 protein, particularly when basal levels of functional p53 are high. However, in TP53-mutated AML, XPO7 drives leukemia proliferation by retaining NPAT, an XPO7-associated protein predominantly expressed in TP53-mutated AML, within the nucleus. NPAT depletion induces genome-wide histone loss, compromises genomic integrity, and triggers replication catastrophe in TP53-mutated AML cells. Notably, the analysis of publicly available AML data sets, primary AML samples, and single-cell intrapatient mRNA profiles further reveals elevated XPO7 and NPAT expression in TP53-mutated AML. Finally, we validate the XPO7-NPAT pathway as a critical driver of leukemia progression in vivo using patient-derived xenograft models of TP53-WT and TP53-mutant AML. Our study delineates key molecular mechanisms underlying TP53-mutated AML pathogenesis and identifies the XPO7-NPAT axis as a critical vulnerability in this refractory leukemia subtype.

Brexucabtagene autoleucel (brexu-cel) is an autologous anti-CD19 chimeric antigen receptor (CAR) T-cell therapy approved for adults with relapsed/refractory (R/R) mantle cell lymphoma (MCL) based on the ZUMA-2 cohort 1 (ClinicalTrials.gov identifier: NCT02601313) study in which brexu-cel demonstrated a 93% objective response rate (ORR) and 67% complete response (CR) rate in patients with R/R MCL and previous BTKi therapy (N = 60). Here, we report the primary results of ZUMA-2 cohort 3 (brexu-cel in patients with BTKi-naive R/R MCL). Adults received brexu-cel at 2 × 106 anti-CD19 CAR T cells per kilogram. The primary end point was ORR assessed by independent radiology review committee (IRRC). As of 26 November 2023, 95 patients were enrolled, and 86 received brexu-cel; median follow-up was 15.5 months. The primary end point was met, with a 91% ORR (95% confidence interval [CI], 82.5-95.9; P< .0001; N = 86) and a CR rate of 73% (95% CI, 62.6-82.2). Estimated 12-month progression-free survival (PFS), duration of response, and overall survival (OS) rates were 75%, 80%, and 90%, respectively. Among 95 enrolled patients, the ORR was 82%, the CR rate was 66%, and the 12-month PFS and OS rates (95% CI) were 73% (62.1-80.8) and 85% (75.6-90.7), respectively. Most patients (88%) experienced treatment-related grade ≥3 adverse events, including 4 treatment-related grade 5 events. Consistent with cohort 1, brexu-cel demonstrated a high ORR and similar safety profile. These results support the continued use of brexu-cel in patients with R/R MCL, and consideration in some patients without previous BTKi therapy who have high-risk disease. This trial was registered at clinicaltrials.gov as #NCT04880434.

Thrombin is generated from prothrombin through cleavage at 2 sites by the enzyme prothrombinase, composed of factor Xa (fXa) and fVa. The affinity of fXa for fVa is low, with assembly and function dependent on phospholipid (PL) membranes. Some snakes have evolved venom versions of fXa that bind to fVa with high affinity and efficiently activate prothrombin in the absence of PL. We created a similar high-affinity, PL-independent human prothrombinase with 17 mutations to human fXa (M17). The increase in affinity enabled cryogenic electron microscopy (cryo-EM) structure determination of M17-prothrombinase to a resolution of 3.3 Å. All protein domains were well resolved in the map, except for the γ-carboxyglutamic acid domain of fXa. The main contacts involve the serine protease and epidermal growth factor-like domain 2 (EGF2) domains of fXa and the A2 and A3 domains of fVa, resulting in the burying of a total surface area of 4900 Å2. The map is of sufficient quality to resolve side-chain interactions, including several key M17 mutations. To aid in the placement of the loop C-terminal to the A2 domain (a2-loop), we solved a high-resolution crystal structure of fXa in complex with a synthetic a2 peptide. The acidic a2-loop interacts with the basic heparin-binding site of fXa, involving a conserved antiparallel β-strand interaction. The M17-prothrombinase structure is compatible with data from biochemical and mutagenesis research and provides important new insights into the assembly and function of the prothrombinase complex.

Aberrant activation of RAS/MAPK signaling limits the clinical efficacy of several targeted therapies in acute myeloid leukemia (AML). In FMS-like tyrosine kinase-3 (FLT3)-mutant AML, the selection of clones harboring heterogeneous RAS mutations drives resistance to FLT3 inhibitors (FLT3i). RAS activation is also associated with resistance to other AML targeted therapies, such as the B-cell lymphoma 2 inhibitor venetoclax. Despite the critical need to inhibit RAS/MAPK signaling in AML, no targeted therapies have demonstrated a clinical benefit in RAS-driven AML. To address this unmet need, we investigated the preclinical activity of RMC-7977, a multiselective inhibitor of GTP-bound active (RAS[ON]) isoforms of mutant and wild-type RAS in AML models. RMC-7977 exhibited potent antiproliferative and proapoptotic activity across AML cell lines with MAPK-activating signaling mutations. In cell line models with acquired FLT3i resistance because of secondary RAS mutations, treatment with RMC-7977 restored sensitivity to FLT3i. Similarly, RMC-7977 effectively reversed resistance to venetoclax in RAS-addicted cell line models with both RAS wild-type and mutant genetic backgrounds. In murine patient-derived xenograft models of RAS-mutant AML, RMC-7977 was well tolerated and significantly suppressed leukemic burden in combination with gilteritinib or venetoclax. Our findings strongly support clinical investigation of broad-spectrum RAS(ON) inhibition in AML to treat and potentially prevent drug resistance because of activated RAS signaling.

Acute myeloid leukemia (AML) carrying chromosomal rearrangements involving the lysine methyltransferase 2A (KMT2A) gene frequently relapse after allogeneic hematopoietic cell transplant (allo-HCT). Pharmacological blockade of the menin-KMT2A interaction disrupts the assembly of oncogenic KMT2A complexes on chromatin, thereby attenuating aberrant self-renewal and inducing myeloid differentiation. We found that beyond this antileukemic mechanism, menin inhibition induced class II transactivator and major histocompatibility complex II (MHC-II) expression in KMT2A-rearranged and NPM1-mutated AML cells in vitro and in vivo. Increased MHC-II expression sensitized AML cells to T-cell-mediated elimination after allo-HCT in mice. Menin inhibition also increased MHC-II expression on primary human AML cells, and enhanced the graft-versus-leukemia (GVL) effect in human xenograft models. Mechanistically, menin inhibition increased expression of multiple human endogenous retroviruses (HERV), leading to consecutive interferon-stimulated gene upregulation and enhanced MHC-II expression. Additionally, menin inhibition directly promoted antitumor effector functions of donor T cells, causing increased tumor necrosis factor-alfa, interferon-gamma, perforin, and granzyme A/B production and cytolytic activity. T-cell exhaustion and menin-KMT2A binding to genes encoding for negative regulators of T-cell activation were reduced by menin inhibition. These findings indicate that menin inhibition enhances the GVL effect via the HERV/MHC-II axis in AML cells and promotes cytotoxicity of donor T cells, which provides a rationale for a clinical trial using menin inhibition as maintenance after allo-HCT.

The Group for Research in Adult Acute Lymphoblastic Leukemia (GRAALL)-2014 trial evaluated an intensive, age-adapted protocol for adults aged 18 to 59 years with Philadelphia chromosome negative acute lymphoblastic leukemia. The trial was motivated by findings from the previous GRAALL-2005 study, which reported excessive toxicity from pediatric-inspired therapy in older patients and no added benefit from allogeneic hematopoietic stem cell transplantation (allo-HSCT) among those with an early favorable response to treatment. Thus, the GRAALL-2014 protocol aimed to reduce treatment-related toxicity in patients aged ≥45 years and to limit allo-HSCT to patients with poor measurable residual disease (MRD) responses. A total of 743 patients were included, and outcomes were compared with those of GRAALL-2005 trial. The GRAALL-2014 study demonstrated reduced early mortality and higher complete remission rates in patients aged ≥45 years. MRD-guided transplantation decisions reduced allo-HSCT indications by ∼50%. Although older patients experienced a higher cumulative incidence of relapse, no significant difference in disease-free survival (DFS) was observed compared with historical cohorts across age subgroups. The overall 4-year DFS was 57.1% (95% confidence interval [CI], 53.4-61.1). Notably, 4-year overall survival improved significantly, from 65.5% (95% CI, 61.7-69.8) to 71.7% (95% CI, 67.7-76.0) in younger patients (P = .031) and from 49.6% (95% CI, 43.5-56.5) to 59.5% (95% CI, 53.5-66.3) in older patients (P = .011). These findings highlight the value of individualized treatment strategies that balance efficacy and safety. Future studies should investigate the integration of immunotherapy to further reduce treatment intensity and improve outcomes. This trial was registered at www.clinicaltrials.gov as #NCT02617004 and #NCT02619630.

The repair of pathological gene variants is an ultimate goal in treating genetic diseases; however, developing distinct therapeutic reagents for each of the numerous variants within a gene may not be scalable. Here, we investigated whether base editing to introduce a gain-of-function variant in blood coagulation factor IX (FIX) can increase FIX activity as a targeted therapeutic approach for hemophilia B. We engineered a G:C to A:T substitution at c.1151 of F9 by cytosine base editing to generate R338Q (the Shanghai F9 variant), which markedly increases coagulation factor activity. An adeno-associated virus vector harboring the base editor converted >60% of the target G:C to A:T and increased FIX activity in HEK293 cells harboring patient-derived F9 variants as well as in knock-in mice carrying a human F9 complementary DNA. Furthermore, administration of lipid nanoparticles containing the base-editor mRNA and guide RNA increased FIX activity in mice. These data indicate that cytosine base editing to generate R338Q in FIX is a broadly applicable genome-editing strategy for hemophilia B with residual FIX activity.

GPRC5D has emerged as a promising therapeutic target in relapsed/refractory multiple myeloma (R/R MM), particularly following progression after B-cell maturation antigen (BCMA)-directed chimeric antigen receptor (CAR) T-cell (CAR-T) therapies. RD118 is a novel CAR-T therapy incorporating a fully human single-domain antibody fragment targeting GPRC5D. In this phase 1 study, 18 R/R patients (17 with MM and 1 with a history of primary plasma cell leukemia) received a single infusion of RD118 at 1.0 × 106, 2.0 × 106, or 3.0 × 106 CAR+ T cells per kg. At a median follow-up of 17.0 months, the overall response rate (ORR) was 94.4%, including 72.2% complete or stringent complete responses. Among the 7 patients previously exposed to BCMA-directed CAR-T therapy, ORR reached 85.7%. Median progression-free survival (PFS) was 18.2 months (95% confidence interval, 14.4 to not estimable), with 12-month PFS and overall survival rates of 82.1% and 93.3%, respectively. Cytokine release syndrome occurred in 88.9% of the patients, primarily grade 1 to 2. One patient developed grade 3 immune effector cell-associated neurotoxicity, which resolved within 72 hours. No cerebellar toxicities or treatment-related deaths were reported. These findings support that RD118 is a highly effective and safe therapeutic option for heavily pretreated R/R MM. This trial was registered at www.clinicaltrials.gov as #NCT05759793 and #NCT05219721.

Myeloid leukemia of Down syndrome (ML-DS) is associated with an excellent prognosis but high treatment-related toxicity and mortality. The Phase 3 Clinical Trial for CPX-351 in ML-DS 2018 aimed to maintain the excellent event-free survival (EFS) achieved in the previous ML-DS 2006 trial while reducing the treatment intensity. Intensity-reduced induction and reinduction therapy with cytarabine and idarubicin with or without etoposide was replaced with CPX-351 (66 U/m2 on 3 days in course 1 and on 2 days in course 2). Risk stratification was based on flow cytometric measurable residual disease (MRD) after first induction. High-risk patients received high-dose cytarabine (3 g/m2 per 12 hour) in consolidation; standard-risk patients received cytarabine at a dose of 1 g/m2 per 12 hour. A total of 35 patients were enrolled until the trial was halted because of an unexpectedly high relapse rate. A per-protocol interim analysis revealed a significantly lower 24-month EFS when compared with the ML-DS 2006 trial (69% vs 90%; P< .001). In contrast with previous studies, most patients who relapsed responded to salvage therapy, leading to a comparable 24-month overall survival of 88% (vs 92%; P = .612). CPX-351 demonstrated a favorable toxicity profile with no treatment-related mortality. Positive MRD by error-corrected GATA1 next-generation sequencing, the presence of trisomy 8 or a complex karyotype were associated with an increased risk for relapse. In conclusion, replacing intensity-reduced induction therapy with CPX-351 in ML-DS led to a significantly lower EFS, highlighting the need for dose optimization to balance the efficacy and toxicity in this sensitive patient population. This trial was registered at https://www.clinicaltrialsregister.eu as EudraCT #2018-002988-25.

Oncogenic growth places great strain and dependence on protein homeostasis (proteostasis). This has made proteostasis pathways attractive therapeutic targets in cancer, but efforts to drug these pathways have yielded disappointing clinical outcomes. One exception is proteasome inhibitors, which are approved for the frontline treatment of multiple myeloma. However, proteasome inhibitors are largely ineffective for the treatment of other cancers at tolerable doses, including acute myeloid leukemia (AML), although reasons for these differences are unknown. Here, we determined that proteasome inhibitors are ineffective in AML due to their inability to disrupt proteostasis. In response to proteasome inhibition, AML cells activated HSF1 and increased autophagic flux to preserve proteostasis. Genetic inactivation of HSF1 sensitized AML cells to proteasome inhibition, marked by accumulation of unfolded protein, activation of the protein kinase R (PKR)-like endoplasmic reticulum kinase (PERK)-mediated integrated stress response, severe reductions in protein synthesis, proliferation and cell survival, and significant slowing of disease progression and extension of survival in vivo. Similarly, combined autophagy and proteasome inhibition suppressed proliferation, synergistically killed human AML cells, and significantly reduced AML burden and extended survival in vivo. Furthermore, autophagy and proteasome inhibition preferentially suppressed protein synthesis and colony formation and induced apoptosis in cells from patients with primary AML, including AML stem/progenitor cells, compared with normal hematopoietic stem/progenitor cells. Combined autophagy and proteasome inhibition activated a terminal integrated stress response, which was surprisingly PKR. These studies unravel how proteostasis pathways are coopted to promote AML growth, progression and drug resistance and reveal that disabling the proteostasis network is a promising strategy to therapeutically target AML.