Introduction: Immunotherapies have significantly advanced the treatment of diffuse large B-cell lymphoma (DLBCL), particularly through the success of anti-CD20 monoclonal antibodies such as rituximab and emerging agents including bispecific CD3/CD20 antibodies and CD19-targeting therapies. However, patients continue to relapse due to naïve and acquired treatment resistance. As immunotherapy options continue to expand, there is a growing need for strategies that can accurately predict patient-specific sensitivities and inform optimal immunotherapy-based combination regimens, especially in the post-approval setting where diverse resistance mechanisms may emerge.

Optim.AI™ is a functional precision medicine platform that has demonstrated clinical utility in predicting effective chemo- and targeted therapy combinations for hematological cancers and sarcomas. To expand its applicability to immunotherapies, we developed Optim.AI™ 2.0, which incorporates immune-tumor co-culture models and high-content imaging for functional evaluation of combination immunotherapy responses.

Methods: Tumor samples and matched peripheral blood were collected from DLBCL patients. Peripheral blood mononuclear cells (PBMCs) were isolated and fluorescently labeled to enable multicellular tracking by high-content imaging and exclusion during tumor cell-specific viability analysis. PBMCs were added to tumor cells at a fixed effector-to-target ratio, and Optim.AI 2.0 combinatorial drug sensitivity testing plates were applied to the co-culture system, with up to 12 FDA-approved drugs, including monoclonal antibodies (rituximab, obinutuzumab), antibody-drug conjugates (polatuzumab), bispecific antibodies (epcoritamab, glofitamab), targeted small-molecule inhibitors (venetoclax, everolimus, zanubrutinib), and cytotoxic chemotherapies (gemcitabine, oxaliplatin, cyclophosphamide, doxorubicin). Tumor cell death was assessed at 48 hours post-treatment using a LIVE/DEAD stain. Fluorescent images were acquired and evaluated using a high-content imaging analysis system and tumor-specific killing was quantified while masking the PBMC population. The resulting data were used as input for Optim.AI™ 2.0 to algorithmically rank and compare sensitivities to combinatorial immunotherapy in DLBCL cells.

Results: Optim.AI™ 2.0 enabled efficient and robust quantification of immune-mediated tumor killing, including antibody-dependent cellular cytotoxicity (ADCC), in ex vivo tumor-immune co-culture models. Distinct top-ranked antibody-based combinations were also identified across treatment naïve and relapsed/refractory DLBCL samples, reflecting the variation in patient response towards a fixed drug set and the need for individualized immunotherapy-based treatment for DLBCL patients. These findings collectively demonstrate the feasibility of pairing Optim.AI™ with an integrated co-culture model and image analysis pipeline to evaluate and rank immunotherapy-based combinations.

Conclusion: This study demonstrates the feasibility of Optim.AI™ 2.0, an enhanced co-culture-based platform which provides a physiologically relevant and scalable approach to functionally evaluate immunotherapy drug sets. With further validation, Optim.AI™ 2.0 holds strong potential to support clinical decision-making and expand the use of immunotherapies in DLBCL.

This content is only available as a PDF.
2025
Sign in via your Institution