Genomic and Single Cell Investigation of MRD Clonality in Genetically Engineered Mouse Models of Multiple Myeloma

Background: Minimal, or measurable residual disease (MRD) is the lowest sensitivity of detection of residual tumor after achieving a meaningful clinical response in multiple myeloma (MM). MRD negativity is associated with deeper response rates and longer progression free survival (PFS). Unfortunately, most patients who achieve MRD negativity eventually relapse, typically with a more aggressive and treatment-resistant malignancy. We set out to develop a preclinical model to study the biology of MRD to a standard of care (SOC) regimen and the relationship to the development of a therapy-resistant MM clone.

Methods: To study MRD and resistance to SOC Revlimid (lenalidomide; R), Velcade (bortezomib; V), and dexamethasone (d), (collectively - RVd) in MM, we utilized genetically engineered mouse models of human-like MM which carry cMYC, and IKK2 expression from germinal-center B cells. Upon development of the disease as determined by bone marrow (BM) aspirates and peripheral gamma protein fraction, mice were treated with RVd therapy for 20 weeks. BM specimens (portional cohort sacrifice) were collected at baseline, 8 weeks (when maximal response was observed, defined as MRD) and >20 weeks (refractory/resistance, RR) and profiled using whole-genome sequencing (WGS) vs. germline control, bulk RNAseq (plasma cells alone and all remaining cells separately) and single cell RNAseq (scRNAseq) of non-sorted flushed BM.

Results: Overall, RVd resistant plasma cells compared to baseline tumors were associated with significantly higher mutation load (median: RVd resistant 16,595; baseline 5,997, p-value = 0.0013, Wilcoxon rank-sum test) and more copy number variations (CNV) including amplifications of whole chromosomes (7, 11, 15, and 17). Further analysis identified two distinct groups of tumors resistant to RVd: one group (CNV-high) is characterized by high genetic instability (CNV-high) including a case of an acquired focal deletion of p53, while the other group (CNV-low) exhibits absence of copy number changes. These patterns were also reflected at the transcriptomic level, with each group exhibiting activation of different pathways, notably lower p53 pathway activity in CNV-high. Despite such differences, both groups showed elevated expression of genes related to TNF−alpha signaling via NF−kB (albeit much lower in the CNV-low), and hypoxia. Analysis of the scRNAseq captured at baseline, MRD and RR was able to identify GFP-expressing CD138+B220- MM cells. Focused differential expression analysis within this specific population revealed a similar increased pathway activation of TNF−alpha signaling via NF−kB, interferon alpha/gamma response, and hypoxia in tumors collected at MRD and RR compared to baseline as observed in bulk plasma cell RNAseq, confirming that these changes are intrinsic to the tumor. For correlation studies, we investigated a cohort of human newly diagnosed (NDMM) and relapse/refractory multiple myeloma (RRMM) RNAseq from CD138 isolated samples and found the upregulation of these pathways in RR compared to NDMM, underscoring the relevance of our findings from the mouse model to human disease.

Conclusion: Our model provides a valuable preclinical tool for investigating the biology of MRD and resistance to a SOC regimen in multiple myeloma. Mice models showed an increase of CNV and mutations at RR as a key characteristic of progression, as well as the activation of TNF-alpha signaling and a more hypoxic environment, which were further validated on patient level data. This analysis enhances our understanding of the underlying mechanisms in a longitudinal manner and potentially guides the development of more effective therapeutic strategies.