CLDN18.2-Directed Therapeutics in Gastric and Gastroesophageal Junction Adenocarcinoma: Biomarker Assessment, Expression Dynamics, and Treatment Sequencing

Claudin-18 isoform 2 (CLDN18.2) has rapidly moved from a gastric-lineage surface antigen to an actionable therapeutic target in advanced gastric and gastroesophageal junction (G/GEJ) adenocarcinoma, following the SPOTLIGHT and GLOW Phase 3 trials, which established zolbetuximab plus fluoropyrimidine-platinum chemotherapy as a first-line option for patients with CLDN18.2-positive, HER2-negative disease. While many existing CLDN18.2 reviews have centered on target biology, assay development, or individual therapeutic platforms, this review focuses on how CLDN18.2-directed therapies can be selected and sequenced within a multi-biomarker landscape in which CLDN18.2 expression is heterogeneous and treatment-modifiable, and its clinical interpretation is modality-dependent. At the same time, the field has expanded beyond conventional monoclonal antibodies to include antibody-drug conjugates (ADCs), bispecific antibodies, and chimeric antigen receptor (CAR) T-cell therapies, each with distinct implications for biomarker threshold requirements, toxicity, and clinical positioning. A central unresolved issue is therefore no longer whether CLDN18.2 can be therapeutically targeted, but how different CLDN18.2-directed modalities should be positioned across treatment lines and after prior CLDN18.2 exposure. This review examines the biologic rationale for CLDN18.2 targeting, current approaches to assay interpretation and patient selection, and the available clinical evidence for the major therapeutic platforms. Particular emphasis is placed on three practice-relevant issues: heterogeneity and sampling in biomarker assessment, treatment-associated modulation of CLDN18.2 expression, and the consequences of these dynamics for post-zolbetuximab decision-making. We also compare next-generation platforms according to mechanistic class, expression-threshold assumptions, toxicity trade-offs, and feasibility of integration into real-world treatment algorithms. Collectively, current evidence supports a reassessment-based rather than assumption-based approach to CLDN18.2 sequencing. However, key gaps remain, including the optimal interface with PD-1-based first-line therapy, the geographic generalizability of several next-generation datasets, standardized retesting strategies at progression, and prospective validation of modality-specific sequencing after target modulation.