Single-Cell Multiomics Analysis Reveals Potential Drivers of Response to the Anti-TIM3 Inhibitor Sabatolimab Combined with Azacitidine in MDS and CMML

Background

The median survival for patients with newly diagnosed (ND) higher risk myelodysplastic syndrome (MDS) and chronic myelomonocytic leukemia (CMML) remains poor. The potential curability of MDS and CMML with allogeneic hematopoietic stem cell transplantation supports the concept of anti-tumor immunity and has led to interest in evaluating immune-based therapeutic approaches in myeloid neoplasms. The immune checkpoint molecule TIM-3 (encoded by the gene HAVCR2) is a target of interest in myeloid neoplasms given its expression on leukemic stem cells as well as several subsets of immune cells including T cells, monocytes, dendritic cells and NK cells. Sabatolimab (MBG453) is an investigational IgG4 anti-TIM-3 antibody currently under evaluation for myeloid neoplasms. In this exploratory study, we sought to characterize the effects of sabatolimab combined with the hypomethylating agent (HMA), azacitidine, on the immune landscape using single cell sequencing of samples from subjects with MDS and CMML treated with the combination from a previously reported phase 1b study (NCT03066648; Brunner et al 2022).

Methods

Following IRB approval, single-cell RNA sequencing (scRNA-seq) and associated proteomic cellular indexing of transcriptomes and epitopes sequencing (CITE-seq) was performed on both blood (BLD) and bone-marrow (BM) derived samples from subjects with: 1) MDS (n=3) or CMML (n=2) treated with HMA therapy alone as part of standard care; 2) ND MDS (n=5) or CMML (n=3) treated with azacitidine combined with sabatolimab; and 3) relapsed/refractory MDS (n=3) treated with sabatolimab alone. Paired BLD and BM samples as well as samples from serial treatment time points were selected whenever feasible (Fig 1). A total of 206,183 cells from BM and 172,421 cells from BLD post-quality control filtering were analyzed. Further, scRNAseq data from normal BLD(200,000 cells) and BM(240,650 cells) generated as part of the human cell atlas (https://data.humancellatlas.org/explore/projects/cc95ff89-2e68-4a08-a234-480eca21ce79) were integrated with this dataset to assist with immune cell subset characterization, enable downstream comparisons to healthy hematopoietic cells, and to help define cell subsets associated with disease states.

Results

In baseline samples from subjects with MDS treated with azacitidine-sabatolimab, we found that an increased baseline abundance of interferon-responsive CD8 T cells in both the BM and BLD (FDR<0.1) was associated with response to therapy. An increased baseline abundance of plasmacytoid dendritic cells and a granulocyte population also suggested an association with response to therapy. Differential gene expression analysis of baseline samples from responding versus non-responding subjects with MDS treated with azacitidine-sabatolimab showed a higher baseline expression of TNF and IFNG in BM CD8 T cells and up-regulated expression of MHC-II machinery (HLA-DRB1, HLA-DPA1, CD74) in myeloid cell subsets in responding subjects (FDR <0.1), suggesting that pre-treatment ability of these cells to be involved in antigen-presentation may play a role in response. In subjects with CMML treated with azacitidine-sabatolimab, we observed dynamic changes in cellular abundances and gene expression when comparing baseline to post-therapy timepoints. Specifically, BM CD8 T cell subsets showed an up-regulation of cytotoxicity genes (GZMA), interferon response genes (IFIT2, IFITM1, IFITM2) and IL32 cytokine post-therapy (FDR <0.1). When evaluating myelomonocytic cell populations, we observed an increase in CD16 monocytes post-treatment. Additionally, dendritic cells and monocytes showed an up-regulation of interferon response genes (IFIT1, OAS1, IFI27). Conversely, we observed a down-regulation of metallothionein genes (MT1E, MT1G, MT2A) as well as a collection of transcription factors (NR4A1, FOSL2, JUN, CEBPB, CEBPD), NF-κB inhibitors (NFKBIZ, NFKBIA) and CXCL8 in the post-treatment samples (FDR <0.1).

Conclusions

Our study provides one of the most comprehensive evaluations of the cellular dynamics of anti-TIM3 immunotherapy in patients to date, allowing for the nomination of novel putative predictive biomarkers of response and identification of potential immunomodulatory mechanisms induced by the combination of sabatolimab with azacitidine in MDS and CMML for further future analysis.