CD19 CAR T Cells Drive a Remodeling of the Immune Microenvironment Associated with T-Cell Dysfunction in Acute Lymphoblastic Leukemia

Introduction Chimeric Antigen Receptor (CAR) T cells induce durable responses in B-cell acute lymphoblastic leukemia (B-ALL). However, a substantial proportion of patients remain refractory or relapse with suboptimal CAR T cell activity. The contribution of the tumor microenvironment to the therapeutic response after CAR T cells remains incompletely understood. We hypothesized that the bone marrow (BM) immunological niche reacts to CAR T cell-mediated inflammation by activating inhibitory pathways and molecules. To verify this hypothesis, we performed a transcriptional analysis of CAR T cells and the BM microenvironment from treated patients and validated findings by flow cytometry. Methods Our group developed anti-CD19 CAR T cells generated with the non-viral Sleeping Beauty (SB) transposon vector from donor T cells in patients relapsed after allogeneic hematopoietic stem cell transplantation (Magnani CF et al, J Clin Invest. 2020 Nov 2; 130(11): 6021-6033). We performed single-cell RNA sequencing and spectral flow cytometry-based analysis of BM-resident immune cells from patients with B-ALL. Data in this study were generated from patients enrolled in the FT01CARCIK Phase I/IIb clinical trial (NCT03389035), in the FT03CARCIK Phase II study (NCT05252403) or who received autologous CAR T cells as part of a Phase I/II trial or commercial cell therapy. Mechanistic insights into the pathways and molecules involved were validated in tumor-bearing humanized mice. Results We analyzed hematological laboratory parameters from patients after treatment and we found that CAR T cells elicit a bystander activation of endogenous immune cells, involving both innate and adaptive immunity, which is actively regulated by a process of immune regulation. To elucidate the cellular and molecular mediators involved in dampening CAR T-cell-mediated immune activation, we performed scRNA-seq of BM samples collected at early time points after treatment (1-2 months). Eighteen sequencing libraries were analyzed, for a total of 71407 high-quality cells. Unbiased clustering of BM cells and infusion products was performed by UMAP embedding. More representative clusters were classified into infusion product, endogenous CD4 and CD8 population, B cells, myeloid cells, pDC, NK, and NK-T cells. We observed profound changes in the microenvironment in response to CAR T-cell-mediated inflammation, and an increase in myeloid cells. Significant enrichment in IFN response, hypoxia, and TGF-β-signaling was associated with the expansion of myeloid-derived suppressor cells (MDSCs) and endogenous exhausted CD8+ T cells, findings validated with a cohort of additional 20 matched BM samples pre- and post-CAR T-cell treatment acquired using spectral flow cytometry. By modeling intercellular communications, we revealed that HIF1α, VEGF, and TGFBR2 are key players in the intercellular communication between CAR T cells and the immune niche, driving widespread T-cell exhaustion, and affecting both CAR T cells and endogenous T cells. Moreover, infusion of anti-CD19 CAR T cells leads to increased accumulation of human MDSCs and exacerbation of a hypoxic environment in HSPC-humanized mice bearing a human tumor. Finally, we also assessed the role of tumor microenvironmental factors on the duration of response (DOR) and found that PD1 expression in endogenous T cells post-treatment was associated with a lack of durable response. Conclusions This study highlights the critical role of the tumor microenvironment on CAR T-cell fate and endogenous immunity in B-ALL. We demonstrate that IFN response, hypoxia, and TGF-β signaling lead to general immune suppression after treatment, resulting in endogenous T-cell exhaustion and compromising CAR T-cell efficacy.