CD22-CAR T cell multiomic features linked to patient outcomes in CD19-CAR resistant large B cell lymphoma

Abstract

Abstract Background: Approximately half of patients with relapsed/refractory large B cell lymphoma (R/R LBCL) experience disease progression after chimeric antigen receptor (CAR) T cell therapy targeting CD19 (CAR19). We investigated CD22 targeting CAR T therapy (CAR22) in patients who progressed after prior CAR19 treatment. Here, we report the molecular determinants of durable remission vs. progression via multi-omics analyses. Methods: We analyzed extended follow-up data from 38 patients with R/R LBCL treated with CAR22 in a single-center academic trial (NCT04088890), with median follow-up of 43.5 months. To identify T cell states associated with clinical response, we performed paired single-cell RNA sequencing (scRNA-seq, 10X), single-cell T cell receptor sequencing (scTCR-seq) and bulk assay for transposase-accessible chromatin sequencing (ATAC-seq) on 70 longitudinally collected samples from 19 representative patients. Results: CAR22 demonstrated notable clinical activity with 68% overall response rate and 53% complete response (CR) rate, as previously reported. Importantly, CR were durable, with those patients achieving CR showing markedly improved outcomes with median progression-free survival of 52 months, median overall survival of 56.1 months, and 75% estimated 3-year survival rate. At data cut-off, no patient achieving a CR had died of their lymphoma. Differential gene expression analysis of T cells in the apheresis material identified unique transcriptional signatures associated with response. In association with enhanced cellular proliferation, the stemness-associated transcription factor T cell factor 7 (TCF7) was upregulated, along with lymphoid enhanced-binding factor 1 (LEF1), both transcription factors implicated in the WNT/β-catenin signaling, as well as several key regulators of T cell stemness and effector memory formation, including FOS, LTB, CCR7, BACH2, NR4A1, SGK1, and interferon signaling (IFITM3). T cell stemness scores (based on combined SELL, CCR7, CD28, IL7R, LEF1 and TCF7 expression) were significantly higher in both CD4+ and CD8+ cells from responders CAR T product samples. ATAC-seq analysis provided orthogonal validation of the single-cell RNA-seq findings by confirming chromatin accessibility patterns associated with clinical response. In clinical responders, CAR T cells demonstrated enriched chromatin accessibility at motifs for key transcription factor families, including TCF/LEF (TCF7, TCF7L1, TCF7L2, LEF1) and NF-κB (RELA, RELB, REL, NFKB1), consistent with a stem-like and less-exhausted phenotype. Elevated activity of TBX21 (encoding T-bet, a master regulator of terminal effector differentiation), IKZF3, FOXP3, IRF9 and RXRA (a nuclear receptor governing lipid metabolism and implicated in T cell exhaustion) was associated with poor clinical response. Through scTCR-seq, we captured 273,565 unique TCR clonotypes across all timepoints. Both responding and non-responding patients exhibited relatively limited TCR repertoire diversity in apheresis material, consistent with the immunosuppressive effects of extensive prior chemotherapy and lymphodepleting conditioning. However, significant differences emerged following ex vivo manufacturing. CAR T products from CR patients demonstrated significantly higher Shannon diversity indices compared to products from PD patients. This enhanced diversity was maintained during peak expansion in vivo but was no longer statistically significant at late timepoints. Responders showed preferential expansion of cytotoxic CD8+ T cell and naïve CD8+ T cell clonotypes, whereas non-responders demonstrated enrichment of regulatory T cell (Treg) clonotypes at peak expansion. Conclusions: Our multi-omics characterization reveals that T cells from patients who progressed after prior CAR19 therapy retain the capacity to mediate durable anti-tumor responses when redirected against CD22. Stemness-associated transcriptional programs and preserved TCR diversity represent critical determinants of CAR22 efficacy. These findings challenge the paradigm that T cells are irreversibly compromised after CAR therapy failure and provide insights for optimizing the ideal T cell phenotypes for CAR22 manufacturing to achieve durable remissions. The durable responses observed in a substantial subset of patients, with 75% 3-year survival among complete responders, demonstrate the significant potential of CAR22 therapy as a salvage approach for patients who progress after CAR19 therapy.