OAB-007: Single-cell multiomic analysis identifies regulatory programs in relapsed/refractory multiple myeloma

Introduction Despite new treatment approaches, in most patients with multiple myeloma (MM) resistant subclones expand, finally resulting in the development of relapsed/refractory disease (RRMM). Recent single-cell transcriptomic (scRNA-seq) studies have improved our understanding of tumor heterogeneity in MM, but our knowledge about the regulatory processes and gene expression signatures in resistant subclones is still limited. Thus, we performed single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq) as well as scRNA-seq from the same MM patient samples and developed a new approach for the analysis of the transcriptome and epigenome of competing subclones. Methods CD138+ bone marrow cells of 11 RRMM patients were longitudinally collected and profiled using 10X Genomics scRNA- and scATAC-seq platform. Pre-processing was performed with CellRanger and downstream analysis using Seurat and ArchR. To determine clones based on copy number variations (CNVs), we leveraged inferCNV for scRNA-seq and developed a sliding window approach adapted from a previously published method by Satpathy et al. for scATAC-seq data. Whole genome sequencing was available to confirm the identified CNV clones. Results We obtained chromatin accessibility profiles of >43000 cells after quality control, establishing the first epigenetic map of RRMM with clearly separated clusters for each patient. Samples from distinct MM subgroups, particularly t(4;14) and t(11;14), clustered together in a pseudo-bulk analysis based on transcription factor activity scores from scATAC. To investigate treatment-induced clonal dynamics we aligned sc chromatin accessibility profiles to clones inferred in scRNA-seq based on CNVs enabling us to track individual tumor subclones over time. An example for clonal adaptation is a patient with two subclones that both responded differently to pomalidomide plus anti-CD38 treatment: while clone 1 showed a higher activity for members of IRF family, ZEB1 and KLF family members were more active in clone 2. Branching evolution was e.g. observed in a patient treated with a MEK/BRAF inhibitor. This patient showed two subclones before treatment, achieved a complete remission and relapsed with one dominant clone. One of the initial subclones could be identified as the precursor of this dominant clone with both showing high activity of NFKB family members, while higher activity of SOX10/15 was only observed upon MEK/BRAF inhibition. An example for combined clonal adaptation and selection was observed in a patient with bi-allelic inactivation of TP53 in one of the two subclones before MCL1 inhibition. Interestingly, this clone was depleted at relapse. Conclusions In summary, our new approach to track individual tumor subclones over time indicates clonal adaptation or selection in virtually all RRMM patients and reveals new insights into regulatory mechanisms underlying treatment-resistance, opening potential avenues for new therapies.