Phase 1 Trial for AZD8359: a CD8-Biased T Cell Engager Targeting STEAP2 in m CRPC

Bispecific T cell engagers, while conceptually powerful, have carried a toxicity profile that constrains their clinical utility. AZD8359, a novel trispecific T cell engager now entering its first-in-human clinical trial, represents a thoughtful attempt to resolve that toxicity problem at the molecular engineering level, and its preclinical data are among the most compelling seen in this class.

The target itself is well chosen. STEAP2 (Six-Transmembrane Epithelial Antigen of the Prostate 2) is a metalloreductase expressed almost exclusively on prostate tissue and markedly overexpressed across every stage of prostate adenocarcinoma. That restricted expression profile is not a minor detail; it is arguably the most important safety feature of the entire program. The history of T cell engagers in solid tumors is littered with on-target, off-tumor toxicities arising from antigens that turn out to be less tissue-restricted than initially assumed. STEAP2’s near-prostate-exclusive distribution dramatically narrows that risk. The STEAP2-focused framework has been building for several years, including the STEAP2-directed CAR-T cell therapy AZD0754, which demonstrated antitumor activity in preclinical bone metastasis models. AZD8359 extends that framework into the TCE space with an entirely new structural logic.

What distinguishes AZD8359 from conventional bispecific T cell engagers is its architecture. Most T cell engagers work by bridging a tumor antigen to CD3, a protein present on all T cells, triggering a broad and indiscriminate immune activation. AZD8359 takes a different approach. Built on the TITAN platform, it is a trispecific molecule that simultaneously targets STEAP2 on the tumor cell, the T cell receptor complex, and, crucially, the CD8 co-receptor. That third binding arm is the key differentiator: by physically grabbing CD8, the molecule selectively recruits cytotoxic CD8+ T cells while largely ignoring CD4+ T helper cells.
The preclinical numbers reflect this clearly: a fourfold increase in maximal CD8+ T cell activation, a 40% reduction in cytokine release in vitro, and more than a tenfold drop in systemic cytokines in animal models, all while maintaining the same level of tumor killing.

Cytokine release syndrome, the most clinically limiting toxicity of T cell engagers, is caused mainly by CD4+ T cells, which flood the bloodstream with inflammatory signalling molecules like IFN-γ and IL-6 when activated. CD8+ T cells, the ones actually responsible for killing tumor cells, contribute far less to this cytokine storm. Conventional TCEs activate both populations simultaneously, making CRS an almost unavoidable side effect. The TITAN format sidesteps this by selectively engaging CD8+ T cells and leaving CD4+ cells largely untouched, targeting the root cause of the problem rather than managing its consequences after the fact.

The in vivo data reinforce this story. In immune-humanized mice bearing STEAP2-positive 22Rv1 tumors in a subcutaneous model, AZD8359 drove complete regressions. More impressively, it achieved the same outcome in a C4-2 intratibial xenograft model, a bone metastasis setting. This matters enormously because the bone marrow microenvironment is notoriously immunosuppressive and poorly penetrated by immune effector cells, and many immunotherapy programs deliberately avoid bone-metastatic preclinical models for precisely this reason. Complete regressions in that context add meaningful translational weight to the preclinical package.

AZD8359 is now entering clinical testing as CRIUS-1, a Phase I/II first-in-human, modular, open-label, multicenter study registered in April 2026.

The broader landscape gives important context. Xaluritamig, an anti-STEAP1 bispecific TCE, has now provided proof-of-concept that T cell engagers can generate meaningful responses in mCRPC even in heavily pre-treated patients. That clinical validation, combined with the persistent CRS burden that has complicated its development, sets the stage perfectly for AZD8359’s entry.

Source.

Clinical trial.