Mechanisms of Response and Resistance to Combination Decitabine and Ipilimumab for Transplant Naïve and Post-Transplant AML/MDS

Treatment of patients with advanced myelodysplastic syndrome (MDS) or acute myeloid leukemia (AML) ineligible for intensive chemotherapy currently is mostly non-curative and therapeutic options for relapse after allogeneic hematopoietic stem cell transplantation (HSCT) rarely induce durable remissions. Novel strategies to address disease relapse are thus needed. Immune checkpoint blockade with CTLA-4 antibodies (ipilimumab) is an emerging concept that demonstrated potent clinical activity in relapsed leukemia cutis post-HSCT. Hypothesizing synergism with therapeutic hypomethylation, the ETCTN/CTEP study 10026 evaluated the safety and efficacy of combination decitabine and ipilimumab treatment in transplant-naïve AML/MDS and post-HSCT AML relapse (NCT02890329). Clinical activity has been encouraging with an overall response rate of 20% (post-HSCT) and 52% (transplant-naïve), however most responses lasted for <6 months, particularly among transplant ineligible patients. To systematically identify determinants of response and resistance, we analyzed bone marrow samples from study participants using bulk and single cell RNA sequencing (scRNA-seq), targeted sequencing, and by quantification of plasma analytes. We obtained 304,961 scRNA-seq profiles from 64 serial bone marrow samples (18 patients with 2-5 samples each). To ensure comparable cell annotation across all samples, we mapped cells to a healthy bone marrow reference. While analysis of screening samples did not reveal differences between 10 responders and 8 non-responders in the composition or transcriptional state of T, NK and predominantly malignant myeloid cells, we observed a higher ratio of T/NK to myeloid cells (2.9 vs. 0.1, p = 0.027) in patients achieving complete remission. For the 8 post-transplant patients, we assessed donor engraftment across cell subsets following deconvolution of donor vs. recipient-derived single cells using expressed single nucleotide polymorphisms (souporcell). While 7-73% myeloid cells in responders were donor-derived, myeloid donor chimerism was <7% in non-responders. This was consistent with analyses of bulk RNA and targeted sequencing, which revealed higher expression of proliferation-associated pathways in non-responders and lower median variant allele frequency of recurrent somatic mutations in responders (14.9% vs. 29.4%, p = 0.029). Together, these results indicated lower disease burden in responders at study entry. Leveraging the study design of a single priming cycle of decitabine preceding combination treatment with ipilimumab, we dissected the pharmacodynamics of decitabine and ipilimumab. Focusing on decitabine, we compared cell type-specific scRNA-seq profiles before and after priming. Gene expression changes were detectable in myeloid subsets with enrichment in pathways such as protein translation, metabolism, and apoptosis. Similarly, plasma analyses showed an increase of soluble IL-8, which is mainly expressed by myeloid cells. Following addition of ipilimumab, gene expression changes were detectable in CD4+ T cells, consistent with T cell differentiation, activation and adhesion, and an increase in regulatory T cells, which we confirmed with staining of CD3 and FOXP3 in bone marrow core biopsies of study participants. Together, decitabine and ipilimumab preferentially acted on myeloid and CD4+ T cells, respectively. Finally, we addressed resistance mechanisms. Clinical responses lacked depth as somatic mutations remained detectable in responders, and recipient-derived single cells demonstrated persistence of leukemic clones among progenitor populations, in line with short remission intervals. Previously acquired bulk RNA-seq data from leukemia cutis cases with durable remission after ipilimumab monotherapy showed transcriptomic evidence of increased infiltration with antigen-experienced resident memory T cells and higher expression of CTLA-4 and FOXP3 than in AML/MDS bone marrow. Together, activity of ipilimumab in marrow involved AML/MDS may be limited by the phenotype of bone marrow-infiltrating T cells. Our studies suggest activity of combined decitabine and ipilimumab is impacted by cellular expression states within the microenvironmental niche of leukemia cells. The inadequate elimination of leukemic progenitors motivates to develop novel immunologic approaches for targeting these cell populations.