Acute Myeloid Leukemia (AML) is an aggressive hematologic malignancy requiring concomitant targeting of critical cellular survival pathways due to resistance and frequent relapse with monotherapies. Venetoclax (VEN), a BCL-2 inhibitor, is one such promising clinical agent best utilized in combination therapies due to transient responses and acquired resistance. Given the involvement of the Rho/ROCK pathway in VEN activity, we combined Rho-associated coiled-coil–containing protein kinase inhibitors (ROCKi))with VEN to achieve superior antileukemic activity. The ROCKi (Fasudil, DJ4, GSK269962A) synergized with VEN to enhance cytotoxicity in both VEN-sensitive and VEN-resistant cell lines in vitro. Among the three ROCKi, GSK269962A (GSK) was best-tolerated in combination with VEN and effectively inhibited leukemia growth across multiple AML cell line-derived xenograft models in vivo. The GSK+VEN combination exhibited additive to synergistic cytotoxicity in primary AML patient cells ex vivo and enhanced antileukemic activity in a patientderived xenograft model. Additionally, the GSK+VEN combination significantly decreased the clonogenicity of primary AML cells, relatively sparing normal cells. Functional assays demonstrated enhanced apoptosis (Annexin V, caspase-3/7), elevated reactive oxygen species, and mitochondrial depolarization in both VENsensitive and VEN-resistant AML cells following combination treatment. Mechanistically, GSK augmented venetoclax responses by downregulating anti-apoptotic proteins (BCL2, MCL1) and inducing pro-apoptotic mediators (NOXA, MCL1 short isoforms), including in VEN-resistant AML cells. Together, these findings across multiple preclinical AML models demonstrate synergistic antileukemic activity and support combining VEN with ROCKi as a promising therapeutic strategy for AML.

TET2 is a commonly mutated gene in hematologic malignancies, including as an initiating event in clonal hematopoiesis (CH). Its mutation alters hematopoietic self-renewal, differentiation, and systemic inflammation responses. TP53 mutations co-occur with TET2 mutations and are also observed in patients with high-risk clonal hematopoiesis and hematologic malignancies. Using a murine model, we found that HSPCs with both mutations initially promoted a myeloproliferative phenotype. Over time these double mutant HSPCs acquire additional genomic alternations, leading to disease progression to acute leukemias including B-ALL. We observed enhanced inflammatory signatures at transformation and identified NLRP1 as a target of TP53 activation. Decreased response to an inflammatory cell death pathway in the setting of TP53 mutation allows cells to tolerate inflammatory stress. This pathway also modifies response to chemotherapies that induce protein translational stalling. Our results identify a hematopoietic stem cell stress response pathway with implications on adaptation to inflammation and chemotherapy tolerance. Significance:TET2 and TP53 mutations co-operate leading to advanced hematologic malignancy. TET2 mutations promote an inflammatory environment and TP53 mutation supports tolerance to this inflammatory stress.

China’s integration of national reforms with institutional redesign has accelerated drug development and access to medicines. As momentum shifts globally, what can other nations learn from this experience?

Frameshift mutations in exon 12 of nucleophosmin 1 (NPM1 mut) are among the most common mutations in acute myeloid leukemia (AML) and have historically been considered favorable-risk in the absence of FLT3-ITD. In the European LeukemiaNet (ELN) 2024 risk-classification for patients treated with hypomethylating agents plus venetoclax (HMA + VEN), NPM1 mut is not considered favorable when co-occurring with signaling gene (SG) mutations (i.e., FLT3-ITD, NRAS, KRAS). However, due to limited numbers in the original analysis, the prognostic impact of SG mutations in NPM1-mutant AML remains unclear. We evaluated the prognostic significance of NPM1 mut with and without SG mutations in two independent cohorts of patients ≥ 60 years with ELN 2024 favorable- or intermediate-risk AML treated with HMA + VEN. Cohort 1 included 322 patients treated in the academic setting. NPM1 mut (n = 61) was associated with a nonsignificantly longer overall survival (OS) compared to NPM1 wild-type (NPM1 wt) (median, 53.05 vs. 17.03 months, p = 0.10). In multivariable analysis (MVA), SG mutations were not independently prognostic within the NPM1 mut subgroup. Cohort 2 included 816 patients from a real-world community-treated cohort. NPM1 mut (n = 124) had a longer OS compared with NPM1 wt (median, 15.3 vs. 14.4 months, p = 0.03). In MVA, NRAS, KRAS, and FLT3-ITD were independent unfavorable prognostic factors; NPM1 mut with, compared to without, SG co-mutation had a shorter OS (median, 9.4 vs. 31.6 months, p = 0.001). These findings suggest SG mutations negate the favorable impact of NPM1 mut in older patients treated with HMA + VEN. Prospective clinical trials are needed to investigate the use of combination therapies to improve outcomes in this high-risk subgroup.

General-purpose large language models (LLMs) are trained on large corpora to acquire broad knowledge, but whether LLMs can replace, or augment, task-specific models is unclear. We evaluated LLMs on three real-world, clinically important tumor genomic interpretation tasks, in order of increasing difficulty: (i) distinguishing tumor from non-tumor mutations (n=34,415 variants), (ii) distinguishing driver from passenger mutations (n=13,469 variants), and (iii) inferring cancer type from tumor sequencing reports across multiple assays and institutions (n=102,791 samples). The best general-purpose LLMs performed as well as the benchmark tailor-made predictor for task (i). Ensembling tailor-made models with zero-shot LLMs improved their performance for tasks (i) and (ii). For task (iii), LLMs outperformed or supplemented tailor-made models on out-of-distribution data. Without fine-tuning, current LLMs already can be useful in clinical genomic interpretation by adding complementary expertise to tailor-made, state-of-the-art predictors.

Single-cell transcriptomics is valuable for uncovering individual cell properties, particularly in heterogeneous systems. However, this technique often results in the reanalysis of many well-characterized cells, increasing costs and diluting rare cell populations. To address this, we develop PIP-seq for Rare-cell Enrichment and Sequencing (PURE-seq). PURE-seq allows direct FACS sorting of cells into PIP-seq reactions, minimizing handling and reducing cell loss. PURE-seq reliably sequences ultrarare cells, with 1 hour of sorting capturing tens of target cells at a rarity of 1 in 1,000,000. Leveraging this extreme sensitivity, we use PURE-seq to isolate and single-cell sequence circulating tumor cells from metastatic melanoma patient blood, obtaining detailed single cancer cell gene expression profiles. Additionally, we use PURE-seq to examine hematopoietic stem and progenitor cells from young, old and middle-aged mice. Transcriptomic analysis identifies Egr1 as a putative master regulator of murine hematopoietic stem and progenitor cell aging, demonstrating PURE-seq's utility as a discovery platform for basic science applications. PURE-seq offers a simple and highly sensitive method for single-cell sequencing ultra-rare cells.

Azacitidine (Aza) plus venetoclax (Ven) is standard treatment for older/unfit patients with newly diagnosed (ND) acute myeloid leukemia (AML). The approved 28-day (D) Ven schedule is associated with prolonged cytopenias, causing frequent dose reductions and cycle delays. Retrospective studies show similar efficacy and reduced toxicity with abbreviated Ven dosing, but prospective data is lacking. We conducted OPTI-AML(NCT03013998), a prospective randomized phase 2 trial comparing 28D Ven (AV28) versus 14D (AV14) with Aza (75mg/m²x7D) for C1-2 in genomically agnostic ND-AML patients ≥60 years. The primary endpoint was complete remission (CR) rate achieved at any time with two cycles of therapy. Between 2023-2025, 169 patients received AV28 (n=83) or AV14 (n=86). CR across two cycles was 49.4% (AV28) versus 43% (AV14); difference of 6.4% [90%CI:-6.1% to 19.0%], not meeting non-inferiority criteria. Patients with NPM1/ IDH2 mutations had higher CR rates with AV28 (60.9% vs. 33.3%), while CR rates were equivalent (45%) for other subgroups. Composite CR rates were 80.7% (AV28) versus 68.6% (AV14) and MRD negativity was similar (77.6% vs. 76.5%). Although AV28 had more frequent treatment interruptions, count recovery after C2, grade ≥3 adverse events and early mortality were similar. In conclusion, the study did not demonstrate non-inferiority of AV14 compared with AV28 during C1-2 in an unselected ND-AML cohort. However, as the confidence interval for the difference covered 0, CR rate for AV28 was not significantly different than AV14. Certain subgroups may benefit from prolonged Ven exposure, but these findings require validation in larger studies, especially as triplet regimens evolve.

Abstract Introduction: Algorithms trained on real-world data aid in tumor genomic prediction tasks, such as identifying cancer driver mutations and inferring cancer type. The extent to which generalist large language models (LLMs) trained on large natural language corpora can replace or supplement such domain-specific algorithms with zero-shot inference is unknown. Methods: We evaluated the zero-shot performance of proprietary (GPT-5, o3-mini, GPT-4o and Claude 3.7 Sonnet), open weight (DeepSeek and Qwen3) and domain-specialized medical (MedGemma) LLMs on three tasks: (i) Distinguishing tumor-somatic mutations from clonal hematopoietic (CH) variants in patients with matched tumor-whole blood profiling (N=37,179 patients; 54,807 samples), (ii) Classifying oncogenic variants using the OncoKB dataset as a positive control (N=10,489 patients; 10,752 samples; 13,470 variants), and (iii) Predicting cancer type from tumor genomic profiles in the multi-institution AACR GENIE dataset (N=97,074 patients; 102,791 samples). Results: Multiple LLMs approached the accuracy of MetaCH, a supervised model for distinguishing somatic tumor mutations from CH variants. o3-mini achieved the highest accuracy for distinguishing oncogenic driver from benign passenger mutations. Among patients with non-small cell lung cancer and mutations in KEAP1, those with VUSs classified as oncogenic by GPT-5 had worse overall survival than those with VUSs classified as benign. GPT-5, o3-mini, and Claude 3.7 Sonnet had accuracy approaching that of a supervised model, GDD-ENS, at classifying 34 cancer types using tumor genomic profiles. Ensemble approaches combining prediction results from GPT-5 and GDD-ENS improved cross-institutional generalizability and performance by an average of 20%. In their reasoning, LLMs discussed clinically relevant genomic features consistent with feature importances from GDD-ENS. Conclusion: Without task-specific training, LLMs achieved performance comparable to specialized supervised models across all tasks. Citation Format: Jennifer Yu, Madison Darmofal, Michele Waters, January Choy, Thinh N. Tran, Chenlian Fu, Leah Morales, Kaicheng U, Ross L. Levine, Nikolaus Schultz, Michael F. Berger, Quaid Morris, Justin Jee. Large language models for tumor genomic interpretation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4100.

Role of 2-HG and KDM3B catalytic activity in KDM3B dependency

Abstract OncoKB™, a precision oncology knowledgebase developed at Memorial Sloan Kettering Cancer Center (MSK), provides expert-reviewed interpretations of the biological function and clinical actionability for >8,000 somatic alterations in 950 cancer-associated genes. It remains the only somatic cancer variant knowledgebase partially recognized by the U.S. FDA. At MSK, OncoKB™ has annotated over 100,000 patient sequencing reports so far and it supports thousands of cBioPortal daily users for their variant interpretation needs. In addition, thousands of users globally are utilizing OncoKB™ for annotating and interpreting cancer variants. OncoKB™ is publicly accessible at www.oncokb.org. It can be used for free for academic research and requires a fee-based license for clinical and commercial use. OncoKB™ classifies alterations based on the level of evidence supporting the alteration as a predictive biomarker of drug response in a specific cancer subtype. To date OncoKB™ includes 58 Level 1 genes as well as MSI-H and TMB-H (included in the FDA drug label), 6 Level 2 genes (included in professional guidelines), 10 Level 3A genes (predictive of response in well-powered clinical studies), 1 Level 4 gene (predictive of response based on compelling biological evidence), and 12 R1/R2 resistance genes. In 2025, OncoKB™ added 3 novel level 1 biomarkers based on the FDA approval of dordaviprone in H3F3A, H3C2, and H3C3 K28M-mutant diffuse midline glioma. OncoKB™ also promoted NPM1 mutations in acute myeloid leukemia and KRAS mutations in low-grade serous ovarian cancer to level 1 based on FDA approvals of revumenib and avutometinib + defactinib, respectively. Multiple EGFR alterations, including exon 19 in-frame insertions and kinase domain duplications, were also elevated to level 1 following FDA approval of datopotamab deruxtecan for EGFR mutant non-small cell lung cancer (NSCLC). NCCN guidelines for pancreatic cancer and NSCLC added erdafitinib for FGFR1/2 fusion-positive and FGFR1/3-mutant tumors, respectively, establishing these as OncoKB™ level 2 biomarkers. Similarly, the NCCN small bowel adenocarcinoma guidelines listed sotorasib and adagrasib for KRAS G12C-mutant tumors, and the NCCN breast cancer guidelines listed neratinib + trastuzumab + fulvestrant for tumors with ERBB2 mutations, also designating level 2 status in these indications. In sum, 3 novel clinically actionable biomarkers and 20 follow-on precision oncology therapies for existing leveled biomarkers were incorporated into OncoKB™ in 2025. OncoKB™ has started annotating germline variants to enable integrated interpretation of paired tumor-normal sequencing results. Other ongoing efforts include updates to diagnostic and prognostic biomarker annotations, a clinical trial matching module, and piloting AI-based curation workflows to expand OncoKB™ annotation for biomarkers detected by whole genome or transcriptome sequencing and immunohistochemistry. Citation Format: Sarah Phillips Suehnholz, Ritika Kundra, Moriah Heller Nissan, Calvin Lu, Nicole Fernandez, Kelly Cavender, Kinisha Gala, Benjamin Preiser, Reshma Ramaiah, John Konecny, Xiang Li, Subhiksha Nandakumar, Kseniya Petrova-Drus, Mark Ewalt, Nikita Mehta, Yonina R. Murciano-Goroff, James Du, Anoop Balakrishnan Rema, Aijazuddin Syed, A. Rose Brannon, Ahmet Dogan, Diana Mandelker, Zsofia K. Stadler, Alexander Drilon, David B. Solit, Ross L. Levine, Maria E. Arcila, Marc Ladanyi, Michael F. Berger, Jianjiong Gao, Nikolaus Schultz, Debyani Chakravarty. OncoKB™, MSK’s precision oncology knowledge base: 2025 updates [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 2498.

Microenvironment remodelling impacts tumour growth and metastasis, but whether remodelling promotes pre-malignant clonal fitness remains unknown. Here, using single-cell RNA-sequencing of the bone-marrow microenvironment in a mouse model of DNMT3A-mutant clonal haematopoiesis (CH), we identify mesenchymal stromal cells (MSCs) in a molecular state of cellular senescence. Elevated bone-marrow MSC senescence is also observed in humans with CH driven by several common somatic mutations. MSC senescence is induced by mutant haematopoietic cells in a contact-independent manner through production of soluble factors including TNF-α and IL-6. These cytokines activate a Stat3-driven pathway that is necessary and sufficient for MSC senescence induction. Genetic or pharmacological depletion of senescent non-haematopoietic cells reduces the burden of CH and delays progression to myeloid neoplasia. Our findings show that microenvironment remodelling modifies pre-malignant clonal fitness and identifies disruption of the crosstalk between pre-malignant cells and their niche as a cancer prevention strategy.

There is a continued need for identification of novel disease drivers of acute myeloid leukemia (AML) as many patients experience relapse and have poor clinical outcomes. Using genomic analyses of a study dataset of paired diagnosis and relapse specimens (n = 59), we identified recurrent downregulation of CCAAT-enhancer binding protein delta (CEBPD) expression at relapse and inferred CEBPD as one of the key regulators of gene transcription in a subset of relapse patients. Three independent public datasets validated downregulation of CEBPD expression at relapse and predicted it as a candidate tumor suppressor gene in AML. To evaluate CEBPD’s tumor suppressor function, we performed complementary loss- and gain-of-function experiments in human AML cell lines OCI-AML2 and OCI-AML5. Consistent with the prediction, knockdown of CEBPD expression led to activation of MAPK signaling and upregulation of downstream effectors cyclin D1 and TNFα expression with concomitant increase in leukemic growth, while CEBPD overexpression resulted in induction of myeloid differentiation marker CD14 expression in the cell lines. Consistent with prior reports, our integrative genomic analyses and azacytidine treatment experiments further suggest a role for DNA methylation in downregulation of CEBPD expression during AML progression. Collectively, our results provide direct functional evidence for a tumor suppressor function of CEBPD in human cell lines and support prior studies implicating its epigenetic silencing in human AML.

ABSTRACT:Acute myeloid leukemia (AML) is a multiclonal disease, existing as a milieu of clones with unique but related genotypes as initiating clones acquire subsequent mutations. However, bulk sequencing cannot fully capture AML clonal architecture or the clonal evolution that occurs as patients undergo therapy. To interrogate clonal evolution, we performed simultaneous single-cell molecular profiling and immunophenotyping on 43 samples from 32 patients with NPM1 (nucleophosmin 1)-mutated AML at different time points in disease progression. Here, we show that diagnosis and relapse AML samples display similar clonal architecture patterns, but signaling mutations drive increased clonal complexity, specifically at relapse, which correlates with overall survival. We uncovered unique genotype-immunophenotype relationships regardless of disease state, suggesting leukemic lineage trajectories can be hard-wired by the mutations present. Analysis of longitudinal samples from patients on front-line AML therapy identified dynamic clonal and immunophenotypic changes consistent with the genotype-immunophenotype relationships we identified.

Supp Table 3: Loci that are differentially hypomethylated in IMF with JAK2V617F mutations when compared to controls

Recurrent mosaic somatic mutations in circulating leukocytes can be frequently found in the aging population. These mutations frequently arise in epigenetic modifier genes like TET2. In the absence of signs of hematologic malignancies this condition is termed clonal hematopoiesis (CH). Although CH has been associated with increased incidence and adverse outcomes in patients with solid tumors, the disease-specific effects and mechanisms by which CH alters solid tumor biology have not been delineated. To develop insights into the interplay between mutant CH clones and epithelial tumor cells we analyzed a cohort of over forty-seven thousand patients who underwent paired blood and tumor sequencing. After correcting for age, sex, ancestry, stage, smoking, and previous treatment history, we identify unique patterns of poor survival associated with specific CH genotypes and solid tumor histologies. Given that the detection of tumor-infiltrating TET2-CH clones is especially associated with poorer outcomes in thyroid cancer patients (HR 2.18, 95%CI 1.18-4.05, p=0.013), we focused on studying the role of CH in thyroid cancer biology. Compared to other CH alleles, TET2-mutant CH is enriched in the tumor microenvironment (TME) across solid tumors and is associated with adverse prognosis specifically in Anaplastic thyroid cancer (ATC) patients. ATC is a clinically aggressive malignancy with a dismal prognosis. Combined BRAF/MEK inhibition offers significant therapeutic benefit in patients with BRAFV600E-mutant ATCs. However, relapses are common and overall survival remains poor. A hallmark of ATC is significant infiltration with myeloid cells, particularly macrophages. ATCs are most common in the aging population, which also has an increased incidence of TET2-mutant CH. These mutant macrophages have been shown to accelerate CH-associated pathophysiology including atherosclerosis. However, the clinical and mechanistic contribution of TET2-mutant clones to the prognosis and/or treatment response in solid tumors has not been elucidated. Using subclonal murine models of Tet2-mutant CH we confirm that mutant macrophages selectively infiltrate mouse BrafV600E-mutant ATC models (58% higher enrichment of Tet2-/- cells in the TME compared to the WT counterparts) and confer resistance to BRAF/MEK inhibition (68-days median survival compared to no mortality after 147 days). Using single-cell CITEseq, we identify that the overexpression of Tgfβ-family ligands by CH macrophages is the driver of this resistance. Importantly, inhibition of Tgfβ signaling or the depletion of mutant-macrophages completely restored the sensitivity to MAPK pathway inhibition. In summary, this work identified a novel actionable resistance mechanism mediated by a clonally driven process within the tumor-immune microenvironment. Pablo Sanchez Vela, Vera Tiedje, Julie L. Yang, Brian R. Untch, Laura Boucai, Aaron J. Stonestrom, Alberto Bueno Costa, Sebastià Franch Expósito, Sebastià Franch Expósito, Avi Srivastava, Marina Kerpelev, Jillian Greenberg, Matthew Wereski, Amanda Kulick, Kevin Chen, Tianyue Qin, Soo-Yeon Im, Anthony R. Martinez Benitez, Raquel Pluvinet, Merve Sahin, Kamal Menghrajani, Gnana P. Krishnamoorthy, Elisa de Stanchina, Ahmet Zehir, Rahul Satija, Jeffrey Knauf, Robert L. Bowman, Manel Esteller, Sean Devlin, Michael F. Berger, Richard P. Koche, Ross L. Levine, James A. Fagin. TET2-mutant clonal hematopoiesis enrichment in the tumor microenvironment is an actionable driver of treatment resistance in solid tumors: Thyroid cancer as a paradigm [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2552.