QLS2401: A Trispecific PSMA/STEAP1/CD3 T‑Cell Engager for mCRPC

QLS2401 is a trispecific T‑cell engager designed for metastatic castration‑resistant prostate cancer (mCRPC). It binds three targets at once: PSMA and STEAP1 on tumor cells, plus CD3 on T cells. This design allows it to physically connect a patient’s T cells to prostate‑cancer cells, triggering T‑cell activation and killing of the tumor.

Both PSMA and STEAP1 are often highly and co‑expressed in mCRPC, which makes dual‑targeting them attractive. By engaging both antigens, QLS2401 aims to increase avidity for tumor cells and reduce the risk of resistance caused by loss of a single antigen such as PSMA or STEAP1 alone. Preclinical data show that QLS2401 induces potent, target‑dependent, T‑cell‑mediated cytotoxicity against prostate‑cancer cell lines and causes tumor regression in xenograft models.

The molecule has been engineered with optimized CD3 affinity and careful control of antigen‑binding valency so that it activates T cells strongly on tumor cells but less so on normal tissues. This “avidity‑driven” activity helps spare cells with low PSMA and STEAP1 expression, which is thought to widen the therapeutic window. In a toxicity study in animal models, QLS2401 was well tolerated, with a Highest Non‑Severely Toxic Dose (HNSTD) significantly higher than that of a xaluritamig analog, suggesting better tolerability at effective doses.

When compared with the xaluritamig analog in preclinical models, QLS2401 shows either superior or at least comparable antitumor efficacy. Xaluritamig (AMG 509) itself is a bispecific STEAP1×CD3 T‑cell engager that has entered clinical development for prostate cancer, so benchmarking against it places QLS2401 in the context of one of the more advanced prostate‑cancer TCE platforms. Taken together, these data position QLS2401 as a next‑generation, dual‑targeting (PSMA/STEAP1) trispecific TCE candidate for mCRPC, with an emphasis on strong efficacy, reduced antigen‑loss escape, and an improved safety margin in non‑clinical models.

Source.