CD19-CAR-T cells have yielded striking results in patients with B-cell malignancies. However, low antigen expression may prevent durable responses in non-Hodgkin lymphoma (NHL) patients. Here, we propose co-targeting CD19/BCMA as a strategy to prevent tumour escape. Our in-house developed CARs, ARI0001 (CD19-CAR) and ARI0002h (BCMA-CAR) have proven safe and efficacious in patients with B-cell malignancies and multiple myeloma, respectively. We hypothesized that the co-transduction of both lentiviral vectors is a feasible strategy to obtain CD19/BCMA dual-targeting CAR-T cells.

We developed the following dual-targeting approaches: 1) Co-transduction (named ARI0003), 2) Pooled single CAR-T cells, 3) Bicistronic CAR-T cells, and 4) Tandem and Loop-CARs composed of two scFvs in one CAR construct. We tested these strategies in pre-clinical models that mimic CD19 high or low densities. For that, we developed a CD19low cell line of NHL and generated Patient-Derived Lymphoma Spheroids (PDLS) from NHL patients that relapsed after treatment with CD19 CAR-T cells.

Bicistronic, Tandem, and Loop CAR-T cells showed impaired CAR expression and effector functions compared to the ARI0003. ARI0003 and Pool achieved a similar efficacy, and both outperformed ARI0001, especially in a setting of low CD19 expression. Notably, ARI0003 resulted in deep tumour remissions in CD19low NHL xenografts after ARI-0001 failure, while the Pool was less effective in eliminating the disease.

Together, these data show promise for the ARI0003 (co-transduction) as an efficient strategy to obtain CD19/BCMA dual CAR-T cells. Thus, ARI0003 was selected as our final clinical candidate for the treatment of patients with NHL.