A Rare Cell State Underpins Diverse Mechanisms of Adaptive Resistance to CAR T-Cell Therapy in B-Cell Acute Lymphoblastic Leukemia

Abstract

The most common pattern of treatment failure of B-cell acute lymphoblastic leukemia (B-ALL) following CD19-targeting chimeric antigen receptor T cell therapy (CAR T), is a measurable residual disease (MRD) negative response followed by relapse. Whilst in a fraction of patients relapse is associated with loss of function mutations &/or deletions of the CD19 locus, a significant proportion of patients relapse with CD19+ disease. We hypothesised that this arises from a rare CAR T tolerant persister population that is able to survive the strong therapeutic pressure and seed relapse. To address this, we investigated the clonal origin and transcriptional dynamics of “CAR T tolerant persister cells” via time resolved clonal tracing of the progeny of individual B-ALL clones, using expressed molecular barcodes in an immunocompetent mouse model of CD19 targeting CAR T. We demonstrate that relapse is a pre-determined fate of rare B-ALL clones, which each adopt diverse, yet clone-specific, transcriptional phenotypes at relapse. Moreover, each diverse relapse phenotype appears to be underpinned by a shared, rare “origin of relapse” cell state that is common to all relapse-fated clones. Our immunocompetent mouse model of murine CAR T cells targeting CD19 (ID3-h41bb-3z) on a syngeneic, BCR::ABL1(p190) driven B-ALL recapitulates the clinical scenario of MRD-negative remission followed by CD19+ relapse. We labelled B-ALL cells with SPLINTR (Fennell & Vassiliadis et al. Nature 2022), a lentiviral lineage tracing reagent comprising expressed molecular barcodes that are read out by single cell RNA sequencing (scRNAseq), to allow dynamic profiling of the clonal origin and transcriptome of individual B-ALL clones over time, across multiple therapeutic timepoints including pre-treatment, remission and relapse. Tumour initiating frequency in this model of B-ALL is high with 23% of transplanted clones able to establish fulminant B-ALL. Only 0.7% of clones, however, were able to persist following CAR-T pressure and seed relapse. In fact, >99% of post-CAR T relapsed disease was comprised by just 3 (0.3%) individual clones and in one mouse disease relapse was driven entirely a single clone. Remarkably, we found that the same clones were able to drive relapse in multiple mice, consistent with a the presence of pre-determined, heritable cell-intrinsic properties capable of causing relapse following CAR T. To further study this phenomenon, we performed scRNAseq on these murine B-ALL samples to examine the transcriptional profiles of these rare, relapse fated clones prior to and following CAR T. Whilst prior to therapy, relapse fated clones were transcriptionally identical to the CAR T sensitive clones, following CAR T we observed that relapse fated clones adopted distinct transcriptional profiles compared to their transcriptome in the absence of therapeutic pressure, suggesting transcriptional adaptation underpins their ability to survive therapy. We noted that each resistant clone had a unique adaptive response and consequent transcriptional signature. Remarkably however, the adaptive response of each clone in different mice was essentially identical, indicative of strong, conserved clone-intrinsic mechanisms of adaptation. Although each CAR T resistant clone adopted disparate transcriptional states in the bulk ALL, we identified a single rare cell state (~1% of the bulk B-ALL) that was shared between all clones, highly suggestive of a common point of origin from which transcriptional adaptation arises. This rare cell state demonstrates metabolic hallmarks of therapy resistance and a distinct developmental transcriptional profile most analogous to the recently characterised preBCR-dependant stage of normal B cell development, raising the prospect that these cells represent an “origin of relapse” cell state. Preliminary data suggests that this rare cell state may also be conserved in human Ph+ B-ALL. In summary, by modelling CD19+ B-ALL relapse following MRD negative remission after CAR T, we have identified a rare cell state that our analysis suggests serves as a common and necessary point from which a transcriptionally diverse range of B-ALL phenotypes can arise and be maintained. We have demonstrated that relapse following CAR T emerges due to clone-specific transcriptional differences from this rare cell state, a phenomenon which may extend to a diverse range of therapies in B-ALL.