PhaseX as a phase 0.9 translational platform to de-risk BiTE immunotherapy through mechanistic patient explant profiling.

Abstract

2615 Background: Bispecific T-cell engagers (BiTEs) redirect cytotoxic T cells to tumor cells, but responses in solid tumors remain heterogeneous due to antigen variability, T-cell exclusion, stromal barriers, and immunosuppressive cues within the tumor microenvironment (TME). PhaseX (patient-derived hydrogel-assisted explants) preserves native tissue architecture and endogenous immune composition ex vivo, enabling mechanistic, patient-specific interrogation of immunotherapies in a physiologically relevant context. We used PhaseX to define BiTE mechanism of action (MoA) and resolve response phenotypes in head and neck squamous cell carcinoma (HNSCC). Methods: Tumor explants from ~25 HNSCC patients were ex-vivo cultured in PhaseX and treated for 6 days with CD3×B7-H3 or CD3×EpCAM BiTEs with matched controls. Supernatants were collected longitudinally for multiplex secretome profiling. At endpoint, explants were dissociated and subjected to flow cytometry, while some explants were fixed and processed to FFPE sections for multiplex immunofluorescence (mIF). Cancer cells were quantified using epithelial markers (PanCK and p60). Tumor response states were classified as tumoricidal (increased tumor-cell CC3 [cleaved caspase-3] with reduced PanCK + /p60 + cancer cell burden) versus tumorostatic (preserved PanCK + /p60 + cancer cell burden with reduced Ki-67 in PanCK + /p60 + cancer cells). Results: Across both CD3×B7-H3 and CD3×EpCAM BiTE-treated explants, PhaseX captured core BiTE MoA: (i) TCR-driven activation with expansion of activated CD8 + T cells (CD137 + ) and enrichment of tumor-reactive CD8 + subsets (CD137 + CD103 + CD39 + ); (ii) immune trafficking and inflammatory programming consistent with IFNγ signaling, evidenced by increased CXCL9/10/11 and coordinated effector cytokine signatures; and (iii) effector execution in tissue, integrating increased cytotoxic mediators (perforin, granulysin) with spatial mIF evidence of enhanced intratumoral CD8+ infiltration. Tumoricidal responders showed elevated tumor-cell CC3 with concomitant loss of PanCK + /p60 + tumor area, whereas tumorostatic responders showed decreased tumor-cell Ki-67 expression with comparatively limited CC3 induction, supporting biologically distinct on-treatment states. Conclusions: PhaseX serves as a “Phase 0.9” translational platform that preserves the TME and resolves BiTE MoA from activation to trafficking to effector execution, while distinguishing tumoricidal and tumorostatic response phenotypes. This framework supports mechanistic deconvolution, biomarker discovery, and patient stratification to de-risk bispecific antibody development.