ESMO Congress 2024 | OncologyPRO

Abstract 1028P

Background

T cell receptors (TCRs) specifically recognize intracellular antigens, indicating unique advantages in solid tumor therapy. Conventional TCR T therapy targets specific antigen presented by fixed human leukocyte antigen (HLA) allele, resulting in limited target patient population. Fully personalized TCR T emerges as an innovative therapy to overcome the shortage but a timely tumor-reactive personalized TCR identification is technically challenging.

Methods

T cell database with single-cell omics data and tumor-reactive/non-reactive tag was constructed. Tumor-reactive T cell Fingerprint was trained by artificial intelligence (AI) neural network. Patient’s personalized tumor-reactive TCR seqs were identified by matching his/her T cells data with the Fingerprint. Three identified TCR seqs were chosen and presented in autologous T cells for each patient. After lymphodepletion, patients received KSX01-TCRT cells at one of two preassigned doses: 5x10⁹ (Dose 1) or 1x10¹⁰ (Dose 2) TCR T cells. Endpoints are safety and preliminary evidence of efficacy.

Results

Currently, 3 patients were enrolled in each dose level. The most common adverse events were neutropenia (Grade 4, 1/6) and decreased white blood cells (Grade 3, 2/6) due to lymphodepleting chemotherapy. No evidence of cytokine release syndrome or immune effector cell-associated neurotoxicity syndrome was observed. No dose-related toxicities were reported. The objective response rate was 33% (1/3) for Dose 1 and 67% (2/3) for Dose 2. The disease control rate was 100% for both dose levels. The manufacturing time from biopsy to lot release was within 48 days, including TCR discovery within only 10 days, which was tremendously shortened compared to previously reported personalized neoTCR T therapy (median time 219 days). Mechanically, TCR T cells were confirmed to infiltrated to tumor lesions and their cell fate trajectory were depicted by single-cell omics of the pre- and post-therapy tumor biopsies.

Conclusions

AI-driven TCR identification as a promising method significantly shortens the drug manufacture time in personalized TCR T therapy. KSX01-TCRT is well tolerated and offers clinical benefits for multiple advanced solid tumors with no limit of HLA restriction or target expression.