Can a CD19-directed bispecific T-cell engager resolve the clinical paradox of simultaneous life-threatening bleeding and thrombosis in multidrug-resistant autoimmune disorders?

BACKGROUND & AIMS:Chimeric antigen receptor (CAR) T cells have shown great potential in hematological cancers, but lack efficacy in solid tumors, highlighting the need for novel strategies. Stimulator of interferon genes (STING) activation was shown to inflame the tumor microenvironment, but combination of STING agonists and CAR-T cells might be limited by detrimental outcomes of T cell-intrinsic STING activation. In this study, we evaluated the potential of combining STING agonists and CAR-T cells in the context of pancreatic cancer. METHODS:We assessed the synergy of CRISPR-Cas9-edited CAR-T cells and the STING agonist diABZI within a T cell exhaustion model in vitro and both xenograft and syngeneic mouse models in vivo. RESULTS:Combination of STING-ablated CAR-T cells and diABZI resulted in enhanced cancer cell killing, increased CAR-T cell proliferation, reduced exhaustion, and expansion of an effector-memory phenotype in vitro. Mechanistically, superior CAR-T cell functionality required genetic ablation of STING in CAR-T cells and was dependent on cancer cell-intrinsic STING signaling on STING-agonistic treatment. Moreover, we identified a synergistic feedback loop comprising the T cell-secreted cytokines interferon-γ and tumor necrosis factor, which prime STING signaling within cancer cells, thereby potentiating the outcomes of cancer cell-intrinsic STING activation in inducing ameliorated CAR-T cell states. Ultimately, we could demonstrate that combination of STING deficient CAR-T cells and diABZI was able to provide enhanced tumor control in both xenograft and syngeneic mouse models. This was accompanied by increased intratumoral CAR-T cell numbers and reprogramming of the tumor microenvironment in vivo. CONCLUSIONS:Our findings suggest that STING deficient CAR-T cells stand to benefit from STING agonists to improve CAR-T cell therapy for immune-deprived cancers such as pancreatic cancer.

ABSTRACT:T-cell-based immunotherapies have revolutionized treatment paradigms in B-cell malignancies, yet their translation to acute myeloid leukemia (AML) has been hindered by a scarcity of tumor-restricted antigens and the risk of on-target off-leukemia toxicity. FLT3 has emerged as a promising therapeutic target with limited expression in healthy hematopoietic tissues. Here, we performed a head-to-head preclinical comparison of an FMS-like tyrosine kinase 3 (FLT3)-directed bispecific T-cell engager (BiTE) molecule and second-generation FLT3-specific chimeric antigen receptor (CAR) T cells. Both approaches induced potent cytotoxicity against AML cell lines and primary patient-derived cells but spared healthy hematopoietic stem and progenitor cells in vitro. Despite similar short-term efficacy, prolonged antigen exposure demonstrated progressive functional decline and metabolic exhaustion; however, CAR T cells maintained cytotoxic capacity and proliferative potential over time. In AML xenograft models, CAR T cells achieved superior tumor control, prolonged survival, and greater T-cell infiltration than BiTE molecule-treated counterparts. Transcriptomic profiling of T cells recovered from the bone marrow further revealed a distinct exhaustion-associated gene signature in samples from mice that had been treated with the FLT3 BiTE molecule. Importantly, provision of CD86-mediated costimulation enhanced antitumor activity of BiTE-redirected T cells in vitro and in vivo. These findings establish FLT3 as a viable and selective immunotherapeutic target in AML and underscore the functional and transcriptional differences between BiTE molecule-redirected T cells and CAR T cells. Moreover, they reveal a critical role for costimulatory signaling in sustaining the efficacy of T-cell-based therapies in vivo, offering a rationale for improving T cell-redirection strategies in myeloid malignancies.

Flow cytometry is an essential part of hematological stepwise diagnostics. It is the gold standard for diagnosing a range of hematological neoplasms. When performed and interpreted correctly, flow cytometry enables complex hematological diagnoses to be made within a very short time - provided that the indication and interpretation are correct.

Bispecific T-cell engagers (BiTE® molecules) have transformed the treatment of B-cell malignancies, yet clinical activity in AML has been modest. Resistance is driven in part by the genetic heterogeneity of AML, most notably TP53 mutations, present in 10-15% of de novo and up to 25% of therapy-related AML. Thus, we hypothesized that TP53 aberrations in AML contribute to cell-intrinsic and extrinsic resistance against T-cell-based immunotherapy. Cytotoxicity against TP53-deleted (DEL) primary AML cells and TP53-knockdown (KD) AML cell lines was reduced in co-cultures with T cells stimulated with the BiTE molecule AMG 330 (CD3×CD33). In addition, T-cell proliferation and proinflammatory cytokine secretion was impaired in co-cultures with TP53 KD cells. Transwell assays identified the secretome of TP53 KD AML cells as a key contributor to the immunosuppressive effects. Proteomic analysis revealed TGF-β1 in TP53 KD co-cultures as a mediator of T-cell suppression. RNA sequencing of T cells co-cultured with TP53 KD cells uncovered a transcriptional shift toward a senescent cell cycle profile. Our data collectively identify the immunosuppressive secretome of TP53-deficient AML as a key barrier to T-cell-engaging immunotherapies, underscoring an unmet clinical need for strategies able to restore T-cell function in TP53 KD AML.

The efficacy of chimeric antigen receptor (CAR) T cell therapy in solid cancers is limited by immunosuppression in the tumour microenvironment (TME). Prostaglandin E2 (PGE2) is a key factor locally inhibiting T cell function. We hypothesized that targeted ablation of PGE2 signalling in CAR T cells may enhance their activity in PGE2-rich solid tumours. Here we generate knockout CAR T cells double deficient for the PGE2 receptors EP2 and EP4 (EP2-/-EP4-/-) by CRISPR-Cas9 engineering. EP2-/-EP4-/- CAR T cells expanded unabatedly in the presence of PGE2. Further, they effectively controlled syngeneic and human xenograft tumour models in vivo, which was accompanied by intratumoural accumulation and persistence of modified T cells. Improved anti-tumour activity was also observed against patient-derived tumour samples from patients with pancreatic ductal adenocarcinoma (PDAC), colorectal (CRC) and neuroendocrine (NET) cancer. Our data uncovers the detrimental impact of PGE2-mediated suppression on CAR T cell efficacy and highlights EP2 and EP4 targeting as a potential strategy.

T cell-based immunotherapy has revolutionized the treatment of B-cell malignancies, yet applying it to acute myeloid leukemia (AML) is challenging due to difficulties in identifying suitable target antigens without on-target off-leukemia toxicity. Prior studies identified FLT3 as a promising target antigen with restricted expression in the healthy hematopoietic compartment1. Here, we evaluate a FLT3-directed BiTE® molecule and 2nd generation FLT3-specific CAR T cells in a preclinical AML model. We hypothesize that positive costimulatory molecules on AML cells enhance BiTE® molecule effectiveness, while CAR T cells may be less dependent due to their built-in costimulatory domain. Cytotoxicity was assessed in cocultures over time using MPFC. The impact of positive costimulation was tested using our established Ba/F3 model system lacking any human costimulatory molecules. On-target off-leukemia toxicity was evaluated in cytotoxicity and CFU assays of pAML cells and healthy bone marrow (hBM). T-cell exhaustion was examined in a longterm culture system with continuous antigen exposure. Lastly, xenograft AML models and RNA sequencing explored differences between the two platforms in vivo. BiTE® molecule and CAR effectively mediated cytotoxicity against various AML cell lines and pAML cells. Overexpression of the costimulatory CD86 on the target cells significantly increased BiTE® molecule-mediated cytotoxicity cells, while CAR-mediated lysis was unaltered. In both scenarios, low E:T-dependent toxicity was observed against hBM. Mixing experiments showed efficient pAML cell lysis with minimal impact on hBM. CFU assays indicated no effect on hematopoiesis of healthy CD34+ stem cells. 28-day cocultures led to T-cell exhaustion in both settings, marked by reduced proliferation, cytokine secretion, metabolic fitness, and cytotoxicity. CAR T cells demonstrated superior antileukemic activity, with enhanced proliferation and splenic homing in an AML xenograft model. RNA sequencing revealed more exhaustion-related markers in BiTE® molecule-redirected T cells, while CAR T cells showed upregulated glycolysis and fatty acid metabolism. Repeating our in vivo xenograft experiment with CD86-overexpressing AML cells supported our hypothesis that costimulation might be a key factor providing a head start to CAR T cells by increasing T-cell proliferation and persistence in vivo. Our in vitro data show similar cytotoxicity and specificity, while xenograft mouse models suggest CAR T cells offer a significant survival advantage over BiTE® molecules. Our data support the hypothesis that positive costimulation integrated within the CAR constructs boosts T-cell activation, homing, and efficacy, thereby delaying T-cell exhaustion in vivo. Future studies are needed to further dissect differences between BiTE® molecule vs CAR T cell-based immunotherapy and identify suitable patient groups. 1Brauchle et al 2020 Lisa Rohrbacher, Daniel Nixdorf, Helena Stadler, Bettina Brauchle, Florian Märkl, Adrian Gottschlich, Gordon Hoffmann, Nora Philipp, Gerulf Hänel, Martin Kirmaier, Anetta Marcinek, Maryam Kazerani, Alica-Joana Emhardt, Giulia Magno, Rebecca L. Goldstein, Sebastian Theurich, Tobias Straub, Sebastian Kobold, Tara Arvedson, Veit L. Bücklein, Marion Subklewe. BiTE® molecule vs CAR-T targeting FLT3 in AML: How positive co-stimulation tips the scale [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 3174.

Despite recent advancements in the treatment of acute myeloid leukemia (AML) the outcome in the relapsed and refractory setting remains poor (Thol, Dohner et al. 2024). While the use of T cell engaging bispecific antibodies (TCE) targeting B cell lineage antigens such as CD19 (Blinatumomab) or CD20 (Epcoritamab, Glofitamab, Mosuenetuzumab) have induced strong and long-lasting response rates in B cell malignancies (Falchi, Vardhana et al. 2023, Liu, Xi et al. 2023), in AML similar progress is yet to be achieved. Early clinical trials of CD33-TCE (JNJ-67571244, AMG330) or CD123-TCE (Vibecotamab) have shown modest clinical activity (response rates ranging between 0 to 16,6%) and a high degree of treatment-emergent adverse events (TEAE) (Ravandi, Bashey et al. 2023, Short, Bachireddy et al. 2023, Narayan, Pierola et al. 2024). In addition, preclinical work suggests possible on-target-off-tumor toxicity of these antibodies towards hematopoietic stem and progenitor cells (HSPC) (Gill, Tasian et al. 2014), overall questioning the ability of complete hematological recovery after TCE treatment targeting CD33 and CD123. In previous work, utilizing high-dimensional single-cell RNA sequencing (scRNA-Seq), we identified novel AML-associated antigens with lower off-tumor expression and no apparent toxicities towards HSPC (Gottschlich, Thomas et al. 2023). Chimeric antigen receptor (CAR) T cells targeting one of our lead candidates - the colony-stimulating factor 1 receptor (CSF1R) - demonstrated strong activity in preclinical models including several genetically distinct patient-derived xenograft (PDX) models, despite an overall lower antigen density on AML blasts compared to hallmark AML-associated target antigens CD33 or CD123. Given the lower antigen density on AML blasts, it remains elusive whether CSF1R can be readily targeted through bispecific T cell engagers. We developed CSF1R targeted TCE (CSF1R-TCB) based on the well-known CrossMAb® Technology in the 2+1 TCB format, used for the FDA-approved CD20-TCB Glofitamab, and demonstrated its efficacy in preclinical in vitro and in vivomodels. Binding of CSF1R-TCB and its anti-tumor activity was tested in co-cultures with T cells or Peripheral Blood Mononuclear Cells (PBMC) and human AML cell lines (Mv4-11, THP-1, OCI-AML3, PL-21) and primary AML blasts, using fluorescence-associated cell sorting (FACS), luciferase bioluminescence readouts or live cell imaging, respectively. In these assays, CSF1R-TCB demonstrated dose-dependent binding to AML cell lines as well as strong in vitro anti-leukemia activity towards AML cell-lines (Effector to Target (E:T) ratio 1:2 p < 0.0001 for both Mv4-11 and THP-1) and primary blasts (E:T 1:1 p < 0.0001). In flow cytometry-based co-culture assay or colony-forming unit (CFU) assays of HSPC and T cells, CSF1R-TCB did not induce relevant lysis of HSPC, while CD33-TCB lead to near complete depletion of HSPC (E:T 2:1 p < 0.0001). Importantly, utilizing a CD34+ cord blood (CB) stem cell-humanized mouse model, CSF1R-TCB showed significantly lower signs of cytokine release and minimal reduction in HSPC counts compared to CD33-TCB, highlighting the favorable expression profile of CSF1R (p < 0.01 for GMCSF, MIP1a; p < 0.05 for IL2, IL6, MCP1 and MIP1b detected in serum 24 hours after therapy). In a xenograft-derived cell line model using luciferase-positive Mv4-11 AML cells (Mv4-11-luc+), treatment with effector T cells and CSF1R-TCB reduced tumor outgrowth and overall tumor progression compared to control-TCB a germline antigen and CD3-binding TCB (p < 0.0001 at day 38 post tumor inoculation). In summary, we could show the safety and efficacy of CSF1R-TCB in preclinical in vitro and in vivo models and demonstrate the superior safety profile of CSF1R-TCB compared to CD33-TCB in CB-humanized mouse models. In cell line-derived xenograft models of AML, CSF1R-TCB induced anti-leukemia activity, warranting further preclinical and clinical investigations.

Background: Ciltacabtagene autoleucel (cilta-cel) targets B-cell maturation antigen (BCMA) in multiple myeloma (MM). Rare, potentially life-threatening late-onset neurotoxicity syndromes associated with BCMA targeting encompass movement and neurocognitive treatment-emergent adverse events (MNTs). Underlying pathomechanisms of this toxicity and their temporal dynamics remain poorly understood. BCMA expression in the caudate nucleus (Van Oekelen et. al., Nat. Med. 2021) hints towards on-target, off-tumor CAR-T cytotoxicity contributing to MNT pathophysiology. However, reliable biomarkers for predicting or monitoring MNT are currently lacking. We previously suggested positron emission tomography (PET) targeting the 18 kDa translocator protein (TSPO) as a surrogate marker of microglial activation and imaging correlate of ICANS in patients following CD19 CAR-T cell therapy (Vinnakota et al., Nat. Cancer, 2024). Aims: To explore potential biomarkers and gain insight into the pathomechanisms of MNT, we analyzed clinical trajectories under multimodal neuromonitoring in patients (pts) with grade 3/4 MNT following cilta-cel. Methods: We report results from two MNT pts, encompassing longitudinal flow cytometry analyses of peripheral blood (PB) and cerebrospinal fluid (CSF), resting-state fMRI (imaging biomarker for functional CNS connectivity), multimodal positron emission tomography (PET) imaging ([18F]Fluoroethyl-L-tyrosine ([18F]FDG-PET), dopamine transporter single-photon emission computed tomography (DaT-SPECT), and PET imaging with a high-affinity ligand binding TSPO, pyrazolo pyrimidine acetamide [18F]DPA-714 as well as immunoassays of inflammatory markers and established serum biomarkers indicative for neuroaxonal injury and astroglial activation (neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP)). Biomarker analyses were carried out at MNT onset and over a period of 2 months. Informed consent was obtained prior to CAR-T treatment in both pts. Results: Both pts (2 males, 68 & 63y/o) received cilta-cel due to refractory MM (3 and 4 prior therapy lines, respectively). Short-term toxicity included max. grade (G)2 cytokine release syndrome (CRS) in both pts. MNT (G3 in both at onset, assessed as Gmax of reported symptoms per CTCAE) was diagnosed on day 48 (pt 1), and day 21 (pt 2) after cilta-cel transfusion. Neurological examination at onset revealed hypokinetic movement disorder with right-sided rigor pronounced to the upper body, as well as psychomotor retardation and dysexecutive syndrome. MNT rapidly progressed to CTCAE G4 in both patients despite therapies including dexamethasone, anakinra, and IVIGs. Intrathecal chemotherapy was subsequently administered without significant effect. It was followed by i.v. cyclophosphamide, which did not result in resolution, but until days 84/92 post-onset has prevented further MNT progression. Flow cytometry demonstrated extensive CAR-T cell expansion at MNT onset (%CAR-T of CD3+ cells in PB: 93.6% & 100%; in CSF: 95.8% & 97.4%), with sustained persistence in both PB and CSF. In both patients, NfL and GFAP showed a marked increase prior to symptom onset compared to CAR-T baseline and remained elevated throughout the MNT course. Routine MRIs at MNT onset were without pathological findings, whereas resting-state fMRI revealed alterations in frontal and striatal neuronal networks at both MNT onset and later timepoints. Multimodal PET at MNT onset showed marked TSPO tracer uptake in the basal ganglia, whereas FDG-PET and DaT-SPECT did not reveal any uptake changes nor evidence of presynaptic dopaminergic neuron loss. Over the clinical course, TSPO-PET showed significant uptake within the basal ganglia at 2 and 4 weeks after onset, followed by a progressive decline over the following two months. By 2 months, FDG-PET showed reduced metabolic activity in the striatum and thalamus, while DaT-SPECT revealed presynaptic dopamine transporter degeneration, indicating evolving neurodegenerative dopaminergic loss. Conclusion: This is the first report to demonstrate in vivo proof of neuroinflammation of the basal ganglia by TSPO-PET imaging and elevated NfL and GFAP levels in the context of cilta-cel induced MNT at symptom onset, preceding neurodegeneration of presynaptic dopamine transporters. Our results significantly contribute to an improved understanding of MNT pathomechanisms and lay the foundation for future potential therapies of these hard-to-manage neurotoxicities.

Introduction Mutations in IDH1 or IDH2 occur in 15-20% of acute myeloid leukemia (AML) cases and promote leukemogenesis via production of the oncometabolite D-2-hydroxyglutarate, which disrupts metabolism and epigenetic regulation. Although treatment options for elderly and unfit AML patients have significantly improved — with many newer therapies now feasible in the outpatient setting — adverse events such as differentiation syndrome (DS) and febrile neutropenia still require careful patient monitoring. In the phase-3 AGILE-study, DS occurred in 14% of AML patients treated with ivosidenib, with a median onset of 19.5 days (range, 3–33), underscoring the importance of close monitoring during the initial weeks of therapy. We conducted a retrospective multicenter study to evaluate the clinical activity of IDH inhibitors in AML with particular focus on DS. Methods Eleven tertiary care centers across Germany participated in the study. Data were collected retrospectively through medical chart review. AML patients who received either enasidenib or ivosidenib outside of a clinical trial were included. Responses were evaluated according to ELN 2022 criteria, and baseline cytogenetics and next generation sequencing were performed per institutional standard and included in the analysis where available. The study was approved by the TUM institutional review board. Results Fifty-six patients were included in the analysis. Most patients (n=43, 77%) carried an IDH1 mutation and were treated with ivosidenib while 13 (23%) received enasidenib for IDH2-mutated AML. The median patient age was 73 years (range, 38-86) at the start of therapy, 64% were male. The median ECOG performance status was 1 (0-3). Most patients had a normal karyotype (n=29, 52%), a complex karyotype was present in 3 patients. Co-mutations were common, the most frequent being ASXL1 (n=16), NPM1 (n=15), and DNMT3A (n=15). Patients harbored a median of 3 co-mutations (range, 1–7). Of all patients, 59% (n=33) were treated in first line (1L), among those 32 (97%) with ivosidenib. Enasidenib was mostly given as monotherapy for relapsed AML (n=8) or as maintenance after transplant (n=4). Azacitidine was the most frequent combination partner (n=35, 63%), while decitabine or low-dose AraC were used infrequently. One patient received ivosidenib, azacitidine and venetoclax as a triplet. Among previously treated patients, 17 had received intensive chemotherapy and 7 patients had been treated with venetoclax-based regimens. The majority of 1L-treated patients (94%) had favorable risk according to ELN 2024. Among those who received intensive therapy in 1L, 53% (n=9) had adverse risk according to ELN 2022. Clinically relevant DS occurred in 16 patients (29%), with a median onset at 11 days (range, 1–33) after therapy. All but one DS case occurred in patients treated with ivosidenib (94%), most of whom (80%) received it in the first-line setting. Notably, three patients developed a second episode of DS following re-exposure to ivosidenib, occurring at day 14 and day 20 (cycle 1), and at day 62 (cycle 2) of therapy. Among those patients, two were treated for relapsed AML. In two patients, DS was managed by withholding ivosidenib alone. Seven patients additionally received steroids; five received a combination of steroids and hydroxyurea (HU), and one patient was treated with HU alone. Data were unavailable for one patient. According to CTCAE v5.0, DS was defined as grade II in 8 patients, III in 5, IV in 1, and V (fatal) in 2 patients. Notably, both fatal DS events occurred in relapsed AML patients. Febrile neutropenia was observed in 29 patients (52%). Excluding patients who received enasidenib as maintenance therapy, 29 patients (56%) achieved a complete remission (CR) or CR with incomplete count recovery after a median of 74 days (31–230). Of those patients, 26 (90%) were treated with ivosidenib (n=6 in 2L), and 3 had received enasidenib. At the time of data cutoff, patients who responded had received a median of 6 cycles (1-26). Conclusions In this real-world cohort, targeted therapy for IDH-mutant AML demonstrated promising clinical activity, similar to randomized controlled clinical trials. DS occurred more frequently than previously reported. While most cases resolved, two were fatal. We observed a broad time range of DS onset, not limited to the first treatment cycle or to one DS episode, highlighting the need for close patient monitoring.

This study investigates the clinical relevance of the gut microbiome at taxonomic and metabolic levels in anti-CD19 chimeric antigen receptor (CAR) T-cell therapy, both in patients and in a preclinical syngeneic tumor model. Patients with B-cell lymphoma treated with CD19 CAR T cells exhibited profound intestinal dysbiosis, exacerbated after CAR T-cell infusion. This dysbiosis was characterized by low bacterial richness, low soluble MAdCAM-1, and loss of Akkermansia species, associated with resistance to therapy. Mechanistically, oral Akkermansia massiliensis supplementation increased CAR T-cell infiltration into the bone marrow, inverted the CD4/CD8 CAR T-cell ratio, favored Tc1 CD8+ T-cell polarization, and promoted the release of tryptophan-derived indole metabolites, leading to better tumor control. The clinical benefit of Akkermansia spp. supplementation was abolished when CAR T cells were genetically deficient in the indole receptor, aryl hydrocarbon receptor (AhR). AhR-agonistic indoles alone failed to replicate the bacterium's anticancer effects. These findings suggest that Akkermansia supplementation could improve CAR T-cell potency in patients with intestinal Akkermansia deficiency. SIGNIFICANCE:B-cell lymphoma patients treated with CAR T cells harbor major gut microbiota perturbations and related metabolism that restrain CAR T-cell therapy. Reprogramming the gut microbiota ecosystem by oral A. massiliensis supplementation induces CAR T-cell niching and Tc1 differentiation in the bone marrow, promoting tumor control in an AhR-dependent manner.

CAR T cells are genetically modified T cells that target specific epitopes. CAR T cell therapy has proven effective in difficult-to-treat B cell cancers and is now expanding into hematology and solid tumors. To date, approved CAR therapies target only two specific epitopes on cancer cells. Identifying more suitable targets is challenged by the lack of truly cancer-specific structures and the potential for on-target off-tumor toxicity. We analyzed gene expression of potential targets in single-cell data from cancer and healthy tissues. Because safety and efficacy can ultimately only be defined clinically, we selected approved and investigational targets for which clinical trail data are available. We generated atlases using >300,000 cells from 48 patients with follicular lymphoma, multiple myeloma, and B-cell acute lymphoblastic leukemia, and integrated over 3 million cells from 35 healthy tissues, harmonizing datasets from over 300 donors. To contextualize findings, we compared target expression patterns with outcome data from clinical trials, linking target profiles to efficacy and toxicity, and ranked 15 investigational targets based on their similarity to approved ones. Target expression did not significantly correlate with reported clinical toxicities in patients undergoing therapy. This may be attributed to the intricate interplay of patient-specific variables, the limited amount of metadata, and the complexity underlying toxicity. Nevertheless, our study serves as a resource for retrospective and prospective target evaluation to improve the safety and efficacy of CAR therapies.

Patient characteristics according to Akkermansia spp. prevalence in stool at baseline.