Radiomic features of CECT and SUVmax of dual-tracer PET/CT reveal PD-L1 spatial heterogeneity in PDAC

Abstract Pancreatic ductal adenocarcinoma (PDAC) is characterized by limited responsiveness to immunotherapy due to spatial heterogeneity of PD-L1 expression. Invasive biopsy-based assessments may fail to capture this heterogeneity. This study aims to evaluate non-invasive imaging biomarkers to predict the expression of PD-L1 in 52 PDAC patients using CECT and F-18 FDG PET/CT and Ga-68 FAPI-04 PET/CT imaging. Intensity histogram and Gray Level Co-occurrence Matrix (GLCM) features were extracted from CECT using IBEX software. PD-L1 expression was validated by immunohistochemistry, RT-PCR, and Western blotting. Correlation between PD-L1 expression, SUVmax of F-18 FDG, and FAPI-04 PET/CT and radiomic CECT features were assessed. PDAC tissues demonstrated multimodal intensity histograms with lower histogram peak height correlated with high PD-L1 expression. A moderate positive correlation was observed between PD-L1 expression and FDG uptake (R 2 = 0.36) and a weaker correlation with FAPI uptake (R 2 = 0.19). This Multi-parametric imaging can enhance immunohistochemistry, aiding in patient stratification for immunotherapy and overcoming the limitations of single-site biopsies.