Acute myeloid leukemia (AML) resides in an immune-rich microenvironment, yet, immune-based therapies have faltered in eliciting durable responses. Bridging this paradox requires a comprehensive understanding of leukemic interactions within the bone marrow microenvironment. We optimized a high-throughput tissue-microarray-based pipeline for high-plex spatial immunofluorescence and mass cytometry imaging on a single slide, capturing immune, tumor, and structural components. Using unbiased clustering on the spatial K function, we unveiled the presence of tertiary lymphoid-like aggregates in bone marrow, which we validated using spatial transcriptomics and an independent proteomics approach. We then found validated TLS signatures predictive of outcomes in AML using an integrated public 480-patient transcriptomic dataset. By harnessing high-plex spatial proteomics, we open the possibility of discovering novel structures and interactions that underpin leukemic immune response. Further, our study's methodologies and resources can be adapted for other bone marrow diseases where decalcification and autofluorescence present challenges.

Abstract Acute myeloid leukemia (AML) is a heterogeneous malignancy of the blood primarily treated with intensive chemotherapy. The allogeneic T-cell antileukemic activity via donor lymphocyte infusions and stem cell transplantation suggests a potential role for checkpoint blockade therapy in AML. While clinical trials employing these treatments have fallen short of expected results, a deeper exploration into the functional states of T cells in AML could bridge this knowledge gap. In this study, we analyzed the polyfunctional activity of T cells in a cohort of patients with relapsed/refractory (RelRef) AML treated on the clinical trial (ClinicalTrials.gov identifier: NCT02397720) of combination therapy using azacitidine and nivolumab (Aza/Nivo). We utilized the single-cell polyfunctional multiplexed immune assay IsoPlexis to evaluate the CD4 and CD8 T cells in peripheral blood and bone marrow samples collected before and after immunotherapy. This revealed at a pseudobulk level that the CD4 T cells exhibited higher functional activity post-immunotherapy (post-IO), suggesting that CD4-directed therapies may play a role in RelRef AML. Additional single-cell analysis revealed significant differences in baseline polyfunctionality in bone marrows of responders as compared with nonresponders for both CD4 and CD8 T cells. Overall, this study highlights the impact of polyfunctional assessment in understanding CD4 and CD8 dynamics in contexts of therapy in AML. Significance: We found T-cell polyfunctionality differs between local and systemic microenvironments. Enhanced variability in proteomic profiles of bone marrow CD4 T cells post-IO suggests their pivotal role in AML treatment response. Single-cell analysis identified novel CD4 and CD8 T-cell functional groups linked to immunotherapy response within the bone marrow.

Baseline correlation analysis of age with individual cytokines and functional groups.

UMAP visualization depicting batch-corrected Seurat defined clusters.

Baseline T-cell functional activity in PB and BM. A, Heat map of T-cell cytokine expression values in PB (n = 20) and BM (n = 16) samples at baseline split by cell subset and scaled by row with clinical characteristic and functional group annotations. ComplexHeatmap was used for data analysis and visualization. B, ComplexHeatmap comparing cytokine expression based on functional groups of CD4 and CD8 cells between TMEs at baseline in the PB and BM. C, Bar plot representation of the significant functional groups from B. Pairwise analysis using ggpaired shows differences based on functional group expression compared between TMEs (P < 0.05).

Interferon gamma (IFNγ) is a critical cytokine known for its diverse roles in immune regulation, inflammation, and tumor surveillance. However, while IFNγ levels were elevated in sera of most newly diagnosed acute myeloid leukemia (AML) patients, its complex interplay in AML remains insufficiently understood. We aim to characterize these complex interactions through comprehensive bulk and single-cell approaches in bone marrow of newly diagnosed AML patients. We identify monocytic AML as having a unique microenvironment characterized by IFNγ producing T and NK cells, high IFNγ signaling, and immunosuppressive features. IFNγ signaling score strongly correlates with venetoclax resistance in primary AML patient cells. Additionally, IFNγ treatment of primary AML patient cells increased venetoclax resistance. Lastly, a parsimonious 47-gene IFNγ score demonstrates robust prognostic value. In summary, our findings suggest that inhibiting IFNγ is a potential treatment strategy to overcoming venetoclax resistance and immune evasion in AML patients.

BackgroundAn inflammatory bone marrow microenvironment contributes to acquired bone marrow failure syndromes. CK0801, an allogeneic T regulatory (Treg) cell therapy product, can potentially interrupt this continuous loop of inflammation and restore hematopoiesis.MethodsIn this phase 1 dose-escalation study of CK0801 Treg cells, we enrolled patients with bone marrow failure syndromes with suboptimal response to their prior therapy to determine the safety and efficacy of this treatment for bone marrow failure syndromes.ResultsWe enrolled nine patients with a median age of 57 years (range, 19 to 74) with an underlying diagnosis of aplastic anemia (n=4), myelofibrosis (n=4), or hypoplastic myelodysplasia (n=1). Patients had a median of three prior therapies for a bone marrow failure syndrome. Starting dose levels of CK0801 were 1 x 106 (n=3), 3 x 106 (n=3), and 10 x 106 (n=3) cells per kg of ideal body weight. No lymphodepletion was administered. CK0801 was administered in the outpatient setting with no infusion reactions, no grade 3 or 4 severe adverse reactions, and no dose-limiting toxicity. At 12 months, CK0801 induced objective responses in three of four patients with myelofibrosis (two had symptom response, one had anemia response, and one had stable disease) and three of four patients with aplastic anemia (three had partial response). Three of four transfusion-dependent patients at baseline achieved transfusion independence. Although the duration of observation was limited at 0.9 to 12 months, there were no observed increases in infections, no transformations to leukemia, and no deaths.ConclusionsIn previously treated patients, CK0801 demonstrated no dose-limiting toxicity and showed evidence of efficacy, providing proof of concept for targeting inflammation as a therapy for bone marrow failure. (Funded by Cellenkos Inc.; Clinicaltrials.gov number, NCT03773393.) In this phase 1 study, the authors evaluated the safety and preliminary efficacy of CK0801 cord blood-derived allogeneic T regulatory cells in patients with bone marrow failure syndromes with suboptimal response to their prior therapy. There were no dose-limiting toxicities observed, and at the 12-month assessment, partial responses and symptom responses were observed in patients with aplastic anemia and myelofibrosis.

Single-cell analysis of IsoPlexis data for CD4 cells (left) and CD8 cells (right). A, UMAP visualization colored by sample location (orange for PB, purple for BM). B, UMAP colored by time point (green for baseline, pink for post-therapy). C, Neighborhood graph depicting differential abundance testing results obtained from miloR. Colors represent the log fold-change between baseline (red) and post-IO (blue) cells. White neighborhoods are nondifferential (FDR 10%). The edges depict links between cells shared by neighborhoods. D, Beeswarm plot of the distribution of neighborhoods by timepoint. E, Beeswarm plot of the distribution of neighborhoods by UMAP-based cluster.

Single-cell cytokine and polyfunctional group analysis at baseline. Heat map of cytokine expression by cluster for A. CD4 T cells. Annotation bars show cluster breakdown by sample location and timepoint. Hierarchical clustering was used to create new polyfunctional groups (CD4-G1-5), which are highlighted by black boxes. B, Boxplots for CD4 T cells show the comparison of complete responders (CR, brown) and nonresponders (NR, blue) at baseline by sample location in these newly defined polyfunctional groups for T cells. C, Heat map of cytokine expression by cluster for CD8 T cells with hierarchical clustering creating new polyfunctional groups (CD8-G1-6). D, Boxplots for CD8 T cells show comparison of CR (brown) and NR (blue) at baseline.

Exploratory analysis of CR and NR patients at baseline. CRs have significantly higher CD4 effector cytokine expression than NRs in the BM. No significant difference in functional group expression from CD8 cells in either PB and BM between CR and NR.

Baseline T cell individual cytokine activity in PB and BM.

Baseline and post-therapy T cell functional activity within PB and BM.

Abstract Comprehensive investigation of CD8+ T cells in acute myeloid leukemia (AML) is essential for developing immunotherapeutic strategies beyond immune checkpoint blockade. Herein, we performed single-cell RNA profiling of CD8+ T cells from 3 healthy bone marrow donors and 23 newly diagnosed (NewlyDx) and 8 relapsed/refractory (RelRef) patients with AML. Cells coexpressing canonical exhaustion markers formed a cluster constituting <1% of all CD8+ T cells. We identified two effector CD8+ T-cell subsets characterized by distinct cytokine and metabolic profiles that were differentially enriched in NewlyDx and RelRef patients. We refined a 25-gene CD8-derived signature correlating with therapy resistance, including genes associated with activation, chemoresistance, and terminal differentiation. Pseudotemporal trajectory analysis supported enrichment of a terminally differentiated state in CD8+ T cells with high CD8-derived signature expression at relapse or refractory disease. Higher expression of the 25-gene CD8 AML signature correlated with poorer outcomes in previously untreated patients with AML, suggesting that the bona fide state of CD8+ T cells and their degree of differentiation are clinically relevant. Immune clonotype tracking revealed more phenotypic transitions in CD8 clonotypes in NewlyDx than in RelRef patients. Furthermore, CD8+ T cells from RelRef patients had a higher degree of clonal hyperexpansion associated with terminal differentiation and higher CD8-derived signature expression. Clonotype-derived antigen prediction revealed that most previously unreported clonotypes were patient-specific, suggesting significant heterogeneity in AML immunogenicity. Thus, immunologic reconstitution in AML is likely to be most successful at earlier disease stages when CD8+ T cells are less differentiated and have greater capacity for clonotype transitions.

Supplementary Table 1. Clinical Characteristics.

Background: Interferon gamma (IFNγ) is a critical cytokine known for its diverse roles in immune regulation, inflammation, and tumor surveillance. Studies have reported the dichotomous nature of IFNγ signaling in both the pathogenesis of cancer and immunotherapy response. However, its complex interplay with acute myeloid leukemia (AML) remains insufficiently understood. Methods: Three independent RNA-seq datasets were integrated to disentangle the IFNγ signaling in AML. We also performed single cell RNA sequencing (scRNA) on 20 newly diagnosed AML patients to discern relative contribution of cells and cellular communications for IFNγ signaling. Results: Single-sample gene set enrichment analysis was used to examine the transcriptional programs linked to IFNγ signaling in 672 newly diagnosed adult AML patients. We observed higher IFNγ signaling score in subgroups of patients including diploid with monocytic differentiation, inv16 core-binding-factor AML, and del7/7q. Notably, sorted CD34 + cells from 17 healthy donors had markedly lower levels of IFNγ signaling scores than did those from AML patients, making it a predominant feature in AML. Additionally, we found significant positive correlations between IFNγ signaling score and expression of its downstream targets HLA class 1 and 2 as well as T cell dysfunction score, T cell exhaustion score, and T cell senescence score, consistent with the fact that chronic IFNγ can drive T cell dysfunction. In scRNA, we observed that AML cells in patients with diploid monocytic AML had the highest expression of IFNγ signaling followed by del7/7q, while the non-monocytic, diploid AML cells had the lowest (Figure 1A). We then explored whether IFNγ signaling is differentially activated across the AML hierarchies and found the spectrum along the hierarchy with the primitive cells displaying significant higher IFNγ signaling than GMP while reaching the highest in Mono-like state. Transcription factor analysis revealed high regulon activity of interferon regulator factors (IRFs) in AML cells, with elevated levels of IRF1 and IRF5 regulons in del7/7q and del5/5q, respectively and elevated IRF8 regulon in diploid AML cells with monocytic differentiation, consistent with its role as a lineage-determinant factor promoting monocytic differentiation. Cell-cell interactions revealed IFNγ from CD8 T cells and NK cells as a top interaction to AML cells. These observations indicate that the activation levels of IFNγ signaling in AML cells are associated with distinct cellular states and hierarchies, and that disparate regulons of IFNγ signaling characterize distinct patient subgroups. The high level of IFNγ signaling in monocytic patient prompted us to assess its correlation with drug response as monocytic subclones are suggested to have inherent resistance to venetoclax. In BEAT-AML ex vivo drug test data, we found a strong positive correlation between IFNγ signaling score and venetoclax resistance, indicating that IFNγ signaling confers venetoclax resistance. This correlation was validated in an independent cohort. However, the IFNγ signaling score did not predict survival outcomes in our bulk cohort. Therefore, we defined a light weighted IFNγ signature to improve the prognostic sensitivity using the least absolute shrinkage and selection operator model. After regression, 47 genes related to survival were retained, forming a parsimonious IFNγ signature. The new parsimonious IFNγ score revealed a tight positive correlation with HLA class 1 and 2 scores. Importantly, this parsimonious IFNγ score was able to predict patient outcomes in our bulk cohort, whereby a higher score predicted worse survival (Figure 1B). Ultimately, these results suggest that the IFNγ pathway activation is associated with resistance to venetoclax-based therapy and can predict patient outcomes independently of known risk factors, making it a promising target for therapeutic intervention. Conclusions: Characterization of inflammation in AML using independent bulk and scRNA profiling led to the identification of novel drug targets and mechanisms of resistance to targeted therapy. We identified monocytic AML as having a unique microenvironment characterized by high IFNγsignaling in AML cells and immunosuppressive features. IFNγ signaling scores correlated strongly with venetoclax resistance. A parsimonious IFNγ gene signature demonstrated robust prognostic value.

Background: Acute myeloid leukemia (AML) is a heterogeneous disease, marked by leukemia stem cells (LSCs), which significantly contribute to treatment resistance and dismal prognosis. While appreciable efforts have been made to study the diversity of functional and phenotypic states of LSCs, integrating these states from inferred single cell profiling has not been explored. Moreover, the spatial distribution and variability of hierarchical leukemic states within diverse cytogenetic groups, and their potential connection to transcriptional profiles, remain understudied, highlighting a gap in our understanding of this complex disease. Methods: To investigate the heterogeneity across different cytogenetics,we performed single cell RNA profiling (scRNA) of bone marrow (BM) mononulear cells from 20 newly diagnosed adult AML patients with diploid (n=7), del5/5q (n=5), del7/7q (n=5) and double deletion (del5/5q & del7/7q; n=3). A total of 111,130 cells passed quality assessment, of which 56,168 (50.5%) cells were identified as AML cells by integrating flow cytometry, immunohistochemistry, copy number variation, and FISH with the scRNA expression profiles. Functional states of AML cells were predicted by a recently proposed hierarchy of AML LSCs representing distinct maturation states (Zeng et al Nat Med 2022). Phenotypical states were annotated by Symphony using a healthy BM and validated in newly sequenced healthy BM samples. We also applied deconvolution scRNA analysis with spatial GeoMx profiling to identify spatial location of primitive versus differentiated AML cells. Results: Differential gene expression and gene set enrichment analysis (GSEA) revealed transcriptional and biological diversity across cytogenetic groups. Del7/7q cells showed high expression of human leukocyte antigen (HLA) genes and CD74, consistent with enrichment for a inflammatory phenotype revealed by GSEA. Del5/5q group demonstrated enrichment in proliferative pathways and heat shock proteins. Some diploid cells showed high expression of monocytic markers, while the others were highly expressing GMP markers, indicating the co-existence of different subclones. Differentiation states of leukemic cells are reflected in the hierarchy of cellular compositions. There is a diverse spectrum of cell types within each cytogenetic group and each patient. However, AML cells from diploid patients were enriched for GMP-like and monocyte-like cells, while phenotypically HSC-like cells showed an incremental trend in the sequence of diploid, del5/5q, del7/7q, and double deletion (Figure 1A). The inferred functional LSC states in our data also showed diverse distribution across cytogenetic groups, with nondiploid cells harboring more LSC. We also evaluated the biological differences of cytogenetic groups within each phenotypical and functional subset. Our data revealed consistent upregulation of proliferation in del5/5q and inflammation in del7/7q with double deletion preserving both features. However, diploid cells were downregulated in almost all. These data suggest cytogenetics as a major driver of ITH even for phenotypically or functionally similar AML cells. By correlating phenotypical axis with LSC states, we found a shared transcriptional program in LSC-like states across phenotypical axis except for Mono-like and DC-like cells, demonstrating the broad distribution of inferred functional LSC spanning the phenotypical axis from primitive to differentiated cells (Figure 1B). This was also observed in other LSC signatures, suggesting that AML cells with LSC capacity exist across the phenotypic of HSC-like cells to more phenotypically differentiated cells. We then performed spatial deconvolution using GeoMx of CD34 + or CD68 + myeloid areas of interest using the phenotypical signature. GeoMx revealed that primitive cells were more likely to be proximal to bone, while more differentiated states were enriched in distal region from the bone, indicating a spatial migration at time of differentiation. Conclusions: ScRNA profiling of newly diagnosed AML revealed the existence of heterogeneous LSC functional states underlying varying degrees of differentiation. We identified cytogenetics as a major source of ITH. Also, primitive cells were likely to localize to bone regions. These findings provide new insights in developing specific therapeutic strategies to improve clinical outcomes.

Abstract Acute Myeloid Leukemia (AML) is a heterogenous disease characterized by immature blasts at different states of differentiation resulting in marked intra-tumor heterogeneity (ITH). Compared to patients with diploid cytogenetics, deletions in chromosome 7/7q (del7/7q), 5/5q (del5/5q) or double deletion (del5/5q & del 7/7q) confer worse outcomes in AML. We conducted single cell RNA (scRNA) profiling of bone marrow mononulear cells (BMMCs) from 20 newly diagnosed adult AML patients with diploid (n=7), del5/5q (n=5), del7/7q (n=5) and double deletion (n=3) to uncover ITH within each group and reveal cellular hierarchies associated with inferred copy number variations (CNV). Diffusion map analysis revealed heterogenous gene expression pattern in AML cells of del7/7q and double deletion patients, compared to diploid and del5/5q patients. CNV-based subgrouping of AML cells revealed significant enrichment of antigen presentation and immune response pathways in cells with inferred del7/7q or double deletion compared to cells without these deletions in the same patients, suggesting a heterogenous immune state in AML cells correlating with inferred genomic copy numbers. To dissect the composition of cellular hierarchies, AML cells were then projected onto an independent healthy BMMC reference (>20,000 cells) and labeled based on transcriptional similarity to healthy hematopoietic cells. AML cells from diploid patients were enriched for GMP-like and monocyte-like cells, while cells from non-diploid patients were enriched in more primitive states (HSC-like and CMP/LMPP-like). Correlating these hierarchal groups with leukemic stem cell (LSC) hierarchies revealed shared transcriptional program in LSC-like states across HSC-like, GMP-like, CLP-like and erythroid-like cells. This suggests that LSC signatures can also be expressed across AML cells and not just primitive cells. These findings were also validated in deconvolution analysis of bulk transcriptomes from three independent newly diagnosed AML cohorts (TCGA, BEAT-AML and MDACC). Spatial deconvolution of 3 patients using whole transcriptomic GeoMx revealed that primitive cells were more likely to be proximal to bone, while more differentiated states were enriched in distal region from the bone. Our findings reveal complex hierarchies in AML cells sharing the LSC expression pattern, while more primitive cells were likely to localize to bone regions suggesting spatial migration of AML cells at time of differentiation. Citation Format: Bofei Wang, Christopher Ly, Fatima Zahra Jelloul, Enes Dasdemir, Guilin Tang, Sonali Jindal, Yulong Chen, Sreyashi Basu, Poonam Desai, Pamella Borges, Preethi Gunaratne, Natalia Baran, Qing Deng, Dapeng Hao, Sean Post, Michael Green Green, Marina Konopleva, Andy Futreal, Padmanee Sharma, Hussein A. Abbas. Leveraging single cell RNA profiling to uncover intra-tumor heterogeneity across cytogenetic subgroups in acute myeloid leukemia [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5927.