Modeling of a hyper-inflammatory hematopoietic niche ex vivo reveals a multi-step process driving CAR T- cell associated myeloid leukemia in multiple myeloma

Abstract

Abstract Introduction: Although prior exposure to mutagenic agents like melphalan and lenalidomide is a known risk factor for post-cytotoxic therapy myeloid neoplasms (MN-pCT) in multiple myeloma (MM), the increased rate of MN-pCTs reported in clinical CAR-T trials suggests that CAR T cell-mediated inflammation may further contribute to this risk, possibly by promoting the progression of pre-existing clonal hematopoiesis (CH) to overt leukemia. Methods and Results: To better dissect the pathogenetic mechanisms leading to CAR T-cell associated MN-pCT in MM, we developed in vitroTP53-knockout (KO) THP-1 cell line models using CRISPR-Cas9. TP53-KO clones showed resistance to melphalan compared to wild type (WT) with an IC50 shift of 1.4-2.1 µM and had a baseline survival disadvantage in drug-free co-culture, which was reversed upon melphalan exposure, with KO rapidly outcompeting WT clones and reaching a clonal dominance of >95% after 14 days of treatment, irrespective of initial seeding ratios. Upon melphalan withdrawal, recovery of WT cells was observed, while a re-introduction of melphalan again selected for KO clones. These data suggest that clonal selection is a key mechanism by which melphalan contributes to MN-pCT pathogenesis. It is clinically supported by prior whole genome sequencing data from our group which detected SBS99, a mutational signature previously linked to melphalan exposure, in bone marrow mononuclear cells of 4/4 patients diagnosed with a MN-pCT at a median latency of 2.9 (0-7.8) months post-CAR T infusion. To simulate cytokine release syndrome ex vivo, we next determined relevant baseline secretion of inflammatory cytokines by CD4⁺ CAR T-cells (cilta-cel, in-house BCMA-directed CAR-Ts) cultured with OPM-2 cells in vitro. These CAR T-secreted cytokines (IL-2, TNF-α, IFN-γ) were added to competition assays and conferred an additional survival benefit to TP53-KO clones which resulted in dominance of KO clones after 30 days of incubation. This indicates that an inflammatory environment acts as a secondary driver to promote CH expansion subsequently to cytotoxic therapy. To expand our findings beyond TP53, peripheral blood mononuclear cells from three CH-positive patients (DNMT3A vs. TP53-mutated) were subsequently seeded in the CoSeedis® 3D ex vivo platform using two different cytokine supplements, either combined IL-2, TNF-α and IFN-γ (1 ng/ml and 10 ng/ml) or single-agent IL-6 (10 ng/ml). Clonal expansion was determined by targeted sequencing and serial variant allele frequency (VAF) estimation complemented by bulk RNA-seq and proteomic analyses. While VAFs showed only modest increase with IL-2, TNF-α and IFN-γ, consistent expansion was noted with IL-6 with an up to ~6-fold clonal expansion relative to baseline after 14 days of ex vivo cultivation. Further analyses are currently ongoing to confirm the differential impact of cytokines on promoting CH expansion after CAR-T, including potential mitigation by the selective anti-IL-6 monoclonal antibody tocilizumab. These analyses will be presented at the meeting together with correlative RNA and proteome data. Discussion and Outlook: Our data indicate a two-step leukemogenic model in which mutagenic and inflammatory stressors act synergistically to drive CH progression to MN-pCT. These findings support the potential of prospective CH screening and advocate for earlier use of CAR T-cells to mitigate the risk of CH-related secondary leukemia in MM.