In vivo evaluation of autologous anti-BCMA CAR-T therapy in PBMC-humanized SDKO mice reveals donor-specific immune responses

Introduction: Autologous CAR-T therapies targeting B-cell maturation antigen (BCMA) have shown remarkable clinical success in treating multiple myeloma (MM). However, accurate prediction of patient-specific efficacy and adverse events - such as cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS) and Immune effector cell associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS) - remains a significant and ongoing challenge. To address the need for personalized screening of CAR-T immunotherapies, we developed and tested a novel humanized mouse model using the NSG-SGM3-IL15-DKO (SDKO) strain, optimized for PBMC engraftment. This model is a new potential preclinical tool to assess individualized CAR-T efficacy and toxicity in vivo . Methods: We evaluated autologous anti-BCMA CAR-T therapy in vivo using five human PBMC donors in the context of an established BCMA + tumor. For this, SDKO mice were injected intravenously with 1×10⁶ MM.1S-luc cells, a model cell line for multiple myeloma. Ten days post-tumor injection, mice were randomized based on their tumor burden and injected with 3×10⁶ PBMCs. Four hours later, mice received either PBS, 5×10⁶ autologous untransduced T cells (UT), or 5×10⁶ autologous anti-BCMA CAR-T cells. Tumor burden and body weight were monitored over 28 days. Blood samples were collected at SD1, SD2, SD5, SD15, SD21 and SD28 to assess human cell engraftment and plasma cytokine levels and tissues retrieved for histopathology. Results: Autologous CAR-T cells induced rapid and potent tumor regression with low toxicity, whereas UT cells failed to control MM.1S-luc tumors across all donors, with tumor progression even more severe than that observed in the PBS group. However, donor-dependent variability was observed: Donors 0935 and 1295 showed reduced CAR-T expansion, and 0935 failed to achieve complete tumor clearance. Transient tumor resurgence occurred in most CAR-T treated mice, except Donor 5263, where tumors were rapidly and permanently eliminated. In contrast, tumors metastasized to bone marrow in PBMC and UT groups. Cytokine profiling revealed both donor-dependent (IL-9, IL-12p70, IL-12p40) and donor-independent (IL-2, IL-4, IL-5, IFNgamma, TNFalpha, IL-13) secretion patterns. Notably, IL-10 exhibited a unique dynamic: a modest spike at SD1 followed by a CAR-T-dependent decline, while PBS and UT groups showed very high IL-10 levels at SD5, suggesting a link to uncontrolled tumor burden and immune dysregulation. Conclusion: Our results demonstrate that the SDKO PBMC humanized mouse model has the potential to be a valuable tool for preclinical CAR-T testing, enabling patient-specific evaluation of efficacy and safety, and paving the way for safer CAR‑T cells in the clinic.