The immunotherapy arsenal against upper gastrointestinal tumours: A comprehensive review of immune checkpoint inhibitors across neoadjuvant, adjuvant, and metastatic settings

Abstract

Immune checkpoint inhibitors (ICIs) targeting the programmed death-1 (PD-1) / programmed death ligand-1 (PD-L1) axis have fundamentally transformed the management of upper gastrointestinal (GI) malignancies. Across oesophageal cancer, gastric/gastroesophageal junction (GEJ) adenocarcinoma, hepatocellular carcinoma (HCC), biliary tract cancer (BTC), and pancreatic ductal adenocarcinoma (PDAC), checkpoint blockade has moved from investigational to standard-of-care — albeit with strikingly different success rates. In the metastatic setting, phase III trials such as KEYNOTE-590, Checkmate 648, and RATIONALE-306 have established anti-PD-1 agents plus chemotherapy as first-line standards for oesophageal squamous cell carcinoma, whereas Checkmate 649, KEYNOTE-859, and ORIENT-16 have done the same for gastric/GEJ cancer. For HCC, atezolizumab plus bevacizumab (IMbrave150) and the STRIDE (Single Tremelimumab Regular Interval Durvalumab) regimen (tremelimumab plus durvalumab, HIMALAYA) represent paradigm-shifting first-line options. In BTC, the dual approvals of durvalumab (TOPAZ-1) and pembrolizumab (KEYNOTE-966) added to gemcitabine-cisplatin have redefined first-line care. In the perioperative space, durvalumab plus the FLOT (5-fluoracil, leucovorin, oxaliplatin, and docetaxel) regimen, evaluated in the MATTERHORN trial, received FDA approval in October 2025 as the first perioperative immunotherapy for resectable gastric/GEJ cancer. Adjuvant nivolumab (Checkmate 577) remains the established standard for resected oesophageal cancer with residual disease after neoadjuvant chemoradiotherapy. Conversely, adjuvant immunotherapy in HCC has been uniformly disappointing, with IMbrave050's initial recurrence-free survival benefit not sustained at extended follow-up and KEYNOTE-937 yielding a negative result. PDAC continues to resist checkpoint blockade in unselected populations, though mismatch repair-deficient tumours, KRAS-targeted vaccines, and novel immune agonists signal cautious optimism. Biomarker-guided patient selection — particularly PD-L1 combined positive score (CPS), microsatellite instability-high (MSI-H)/deficient mismatch repair (dMMR) status, and HER2( human epidermal growth factor receptor 2) expression — remains pivotal for maximising therapeutic benefit. Importantly, several phase III ICI trials have yielded negative results, including KEYNOTE-061, KEYNOTE-062, JAVELIN Gastric 100, JAVELIN Gastric 300, KEYNOTE-585, ATTRACTION-5, and IMbrave050, underscoring that not all immunotherapy strategies succeed and that careful patient and regimen selection is essential.