Treatment-Induced PD-L1 Dynamics and Temporal Heterogeneity in Locally Advanced and Metastatic Gastroesophageal...

Background/Objectives: Standard guidelines for treating locally advanced and metastatic gastroesophageal cancer (GEC) involve combination regimens of chemotherapy and immunotherapy. Patient eligibility for these treatments is primarily guided by the programmed cell death ligand 1 (PD-L1) combined positive score (CPS). This approach assumes that a single baseline biopsy accurately reflects a stable, tumor-wide PD-L1 status, an assumption increasingly challenged by spatial and temporal heterogeneity. This systematic review aims to synthesize existing data on PD-L1 expression dynamics before and after systemic therapy in GEC, and to evaluate whether these changes differ between neoadjuvant and metastatic settings. Methods: We systematically searched PubMed/MEDLINE (up to May 2026) for studies reporting paired pre- and post-treatment PD-L1 status in gastric and gastroesophageal junction cancers. This systematic review was conducted in accordance with the PRISMA 2020 guidelines. Narrative synthesis and qualitative data transformation were structured following the Synthesis Without Meta-analysis (SWiM) reporting framework. Study-level risk of bias was assessed using the Newcastle-Ottawa Scale (NOS). Results: Eleven studies were included in this systematic review, stratified into neoadjuvant (n = 6) and metastatic (n = 5) settings. Our findings suggest that changes in PD-L1 expression in gastroesophageal cancer depend heavily on the clinical setting. Specifically, an upward trend in post-treatment PD-L1 expression was evident in the neoadjuvant setting, with 5 of 6 studies reporting a prevalence of PD-L1 gain ranging from 6.3% to 33.3%. Conversely, PD-L1 downregulation was consistently reported across all metastatic studies, with loss rates ranging from 19.3% to 44.4%. Conclusions: PD-L1 status in gastroesophageal cancer changes dynamically after treatment. These findings highlight the limitations of relying on archival tissue and support the use of repeat biopsies to track evolving tumor phenotypes. This study is primarily limited by small sample sizes of paired biopsies, technical heterogeneity in PD-L1 scoring and treatment regimens, potential publication bias, and the lack of prospective registration.