Immune exhaustion limits CD19 CAR-T efficacy in Richter's transformation Free

Background: Despite recent advancements in the treatment of chronic lymphocytic leukemia (CLL), Richter's transformation (RT) into diffuse large B-cell lymphoma (DLBCL) remains a therapeutic challenge with poor outcomes (Coombs et al., 2023). While CD19-directed chimeric antigen receptor T-cell therapy (CAR-T) has shown efficacy in de novo DLBCL, its effectiveness in RT is less well defined (Kittai et al., 2024). Factors such as prior treatment, T cell exhaustion, and distinct immune microenvironments may contribute to limited efficacy in DLBCL (Fraietta et al., 2018; Jain et al., 2023). We evaluated clinical outcomes and immune correlates of CD19 CAR-T in patients with RT at our institution.

Methods: We retrospectively analyzed 15 patients with RT-DLBCL treated with CD19 CAR-T between 2017 and 2024. A comparator cohort of 19 patients with de novo DLBCL treated with CD19 CAR-T was also analyzed. Pre-infusion peripheral blood mononuclear cells (PBMCs) were analyzed using multiparametric flow cytometric profiling. RNA expression from tumor biopsies was analyzed using NanoString transcriptomic profiling with the IO360 panel, which includes 770 genes found in the tumor microenvironment.

Results: Fifteen patients with RT-DLBCL were included, 47% male with a median age at infusion of 69 years (range 47-77). The median interval from CLL to RT diagnosis was 41 months (range: 7-210), and from RT diagnosis to CAR-T was 20 months (range: 2-50). CAR-T products used were tisagenlecleucel (tisa-cel, 53%), lisocabtagene maraleucel (liso-cel, 27%), and axicabtagene ciloleucel (Axi-cel, 20%). One patient had received an allogeneic CAR-T product 7 months prior to Tisa-cel. Three patients received hematopoietic stem cell transplant before CAR-T (2 autologous, 1 allogeneic). Leading up to CAR-T, 93% had received RT-directed therapy (median 4 lines) and 66% had received prior CLL therapy. Cytokine release syndrome (CRS) occurred in 67% (7% grade ≥3) and immune effector cell-associated neurotoxicity syndrome (ICANS) in 40% (13% grade ≥3). Overall response rates (ORR) were 80% at day 30 and 47% at 3 months. The median follow-up was 9 months.

In the comparator de novo DLBCL cohort, median age at CAR-T was 68 (25–76), with 68% male. Most (74%) had received 1–2 prior lines of therapy before CAR-T. The majority (18/19) received axi-cel, and one patient received tisa-cel. Three patients received an autologous stem cell transplant before CAR-T. CRS of any grade occurred in 79% (11% grade ≥3), and ICANS in 74% (42% grade ≥3). The ORR was 88% at day 30 and 82% at 3 months. The median follow-up period was 26 months (range: 1-67).

Flow cytometry revealed significantly higher frequencies of exhausted T cells in RT patients compared to de novo DLBCL (p=0.0329). RT patients had significantly higher frequencies of TCF1⁺ PD1⁺ (p=0.0217) and TCF1⁺ TIGIT⁺ (p=0.0158) CD8⁺ T cells, indicating a predominance of dysfunctional precursor-exhausted populations in circulation prior to CAR-T infusion. Moreover, there was an increase in central memory (CM, CD45RO⁺ CCR7⁺) CD8⁺ T cells in RT samples compared to de novo DLBCL (p=0.0436). Notably, these CM cells exhibited a significantly more exhausted phenotype in RT patients than in de novo cases (PD-1+ p=0.0266, TIGIT+ p=0.0490). Supporting the exhaustion phenotype found by flow cytometry, Nanostring analysis confirmed a significant increase in ENTPD1 expression, a gene that encodes the ectonucleotidase CD39, in tumors from RT patients compared to de novo DLBCL (p=0.02), as well as an increase in DUSP2 expression (p<0.0001), a phosphatase associated with exhaustion on tumor-infiltrating T cells and poor outcomes (Lu 2020; Dong 2018).

Conclusions: CD19 CAR-T therapy shows promising initial responses in RT, though durability remains limited. Compared to de novo DLBCL, RT patients exhibit higher pre-treatment T cell exhaustion and more immunosuppressive tumor gene expression profiles, which may underlie poorer CAR-T durability. Ongoing transcriptomic analyses will further elucidate tumor-intrinsic and microenvironmental mechanisms of CAR-T resistance in RT.