Dynamic circulating biomarkers for predicting adverse events in PRaG therapy: a pooled analysis and model construction from PRaG series clinical trials

BACKGROUND: PRaG therapy (PD-1/PD-L1 inhibitor combined with Radiotherapy and Granulocyte-macrophage colony-stimulating factor) has emerged as a novel, intensive, and promising combined regimen for advanced solid tumors. It has been adopted as a National Health Commission-promoted technology in China, demonstrating significant survival benefits. This study aims to delineate the adverse events (AEs) profile of PRaG therapy and evaluate the predictive value of dynamic circulating biomarkers for moderate-to-severe AEs. METHODS: 203 patients with advanced solid tumors initiating PRaG therapy were prospectively enrolled and stratified by maximum treatment-related AEs severity (AE ≤ Grade1 vs. AE ≥ Grade2). The predictive value was assessed by comparing circulating indicators at baseline and one cycle prior to AEs onset (pre-AEs cycle). A predictive model was then constructed and subjected to independent internal validation. RESULTS: Patients experiencing AE ≥ Grade2 exhibited significantly superior survival outcomes. Multivariate analysis of dynamic circulating indicators revealed that pre-AEs cycle levels of Interleukin-2 (IL-2, p = 0.004), CD4 + effector memory T cells (p = 0.001), CD8 + effector memory T cells (p = 0.012), and the CD4 + effector/CD8 + effector memory T-cell ratio (p < 0.001) were independent predictors for AE ≥ Grade2. Integrating these fine typing of circulating T lymphocyte subsets into the prediction model significantly enhanced performance (AUC increased from 0.685 to 0.867), with independent internal validation confirming model reliability (AUC = 0.913). CONCLUSION: This study successfully delineated the real-world safety profile of PRaG therapy and identified a distinct set of dynamic circulating biomarkers strongly associated with moderate-to-severe AEs. The established non-invasive prediction model offers an effective tool for clinical early warning and proactive management in PRaG patients.