RESUMEN
BACKGROUND: PLAnning Treatment For Oesophago-gastric Cancer: a Randomised Maintenance Therapy Trial (PLATFORM) is an adaptive phase II study assessing the role of maintenance therapies in advanced oesophago-gastric (OG) adenocarcinoma. We evaluated the role of the anti-programmed death-ligand 1 (PD-L1) inhibitor durvalumab in these patients. PATIENTS AND METHODS: Patients with human epidermal growth factor receptor 2-negative locally advanced or metastatic OG adenocarcinoma with disease control or response to 18 weeks of platinum-based first-line chemotherapy were randomised to active surveillance or maintenance durvalumab. The primary endpoint was progression-free survival (PFS). Safety was assessed in all patients who had commenced surveillance visits or received at least one dose of durvalumab. Exploratory survival analyses according to PD-L1 Combined Positive Score (CPS) and immune (biomarker-positive) or angiogenesis dominant (biomarker-negative) tumour microenvironment (TME) phenotypes were conducted. RESULTS: Between March 2015 and April 2020, 205 patients were randomised to surveillance (n = 100) and durvalumab (n = 105). No significant differences were seen in PFS [hazard ratio (HR) 0.84, P = 0.13] and overall survival (OS; HR 0.98, P = 0.45) between surveillance and durvalumab. Five patients randomised to durvalumab demonstrated incremental radiological responses compared with none with surveillance. Treatment-related adverse events occurred in 77 (76.2%) durvalumab-assigned patients. A favourable effect in OS with durvalumab over surveillance in CPS ≥5 and immune biomarker-positive patients was observed compared with CPS <5 and biomarker-negative subgroups, respectively: CPS ≥5 versus <5: HR 0.63, 95% confidence interval (CI) 0.32-1.22 versus HR 0.93, 95% CI 0.44-1.96; biomarker-positive versus negative: HR 0.60, 95% CI 0.29-1.23 versus HR 0.84, 95% CI 0.42-1.65. CONCLUSION: Maintenance durvalumab does not improve PFS in patients with OG adenocarcinoma who respond to first-line chemotherapy but induced incremental radiological responses in a subset of patients. TME characterisation could refine patient selection for anti-PD-L1 therapy above PD-L1 CPS alone.