Abstract
Neoadjuvant therapy (NAT) is the standard of care for borderline-resectable and locally advanced pancreatic ductal adenocarcinoma (PDAC). It can be used to treat resectable PDAC. This study aimed to investigate how NAT remodels the tumor immune microenvironment (TIME) and whether this remodeling translates into survival benefits. We performed spatial and digital pathology analysis of 27 upfront resection patients (naïve group) and 39 age-, gender-, and stage-matched patients who had surgery after NAT (NAT group). AI-assisted digital pathology was used to annotate cancer cells and CD8 + T lymphocytes. Spatial correlation between CD8 + T lymphocytes and cancer cells for each case was assessed using spatial point pattern analysis, followed by generalized linear modeling (GLM) of quadrat counts of CD8 + T cells, with the quadrat counts of cancer cells as the independent variable. The regression coefficient was used to quantify the strength of their spatial correlation and then further assessed for association with patient survival. The analyses showed that the NAT group, compared with the naïve group, had increased spatial correlation of CD8 + T cells with cancer cells, suggesting enhanced effector T cell-cancer cell engagement in the NAT patients. Additionally, patients with a higher degree of spatial correlation between the two cells showed improved after-surgery survival. Through a new methodological framework that takes advantage of AI-assisted digital pathology and spatial point pattern analysis, our study has successfully captured the subtle effects of NAT-induced TIME remodeling and assessed its impact on prognosis of PDAC patients.