Introduction: Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer-related mortality in the United States. Translational studies using human tissue are essential for therapeutic development but are limited by low neoplastic cellularity and the frequent reliance on small biopsy specimens, particularly in advanced disease. To maximize biological insight from limited material, we developed a high-plex spatial proteomics assay tailored to PDAC and an integrated computational workflow optimized for rapid, correlative analyses.

Methods: PULSAR (Pancreatic cancer Unified Landscape Spatial Architecture Report) is a 61-plex protein panel designed for the PhenoCycler platform (Akoya Biosciences), enabling single-cell, spatially resolved characterization of tumor subtype (basal/classical), driver-associated markers, immune infiltration and checkpoint expression, stromal fibroblast heterogeneity, and proliferative status. The panel includes 21 pre-conjugated and 40 custom-conjugated antibodies selected for PDAC relevance and optimized in primary specimens. Image data were processed using machine learning–based tissue and cell segmentation in QuPath, followed by automated cell phenotyping, feature extraction, and visualization through CELLestial, a custom end-to-end analytical pipeline.

Results: In 30 primary PDAC specimens, mean tissue processing and image acquisition time was 63 hours, with 100% analyzable images. Per-marker staining success was 90%, and 90% of cases achieved >90% marker performance. Images averaged 58.7 GB, with 510,474–957,964 cells phenotyped per case. Computational processing required approximately 2 days per cohort. Over 75 cell phenotypes and >1500 specimen-level spatial features were generated.

Conclusion: This integrated high-plex spatial framework enables comprehensive, rapid profiling of PDAC and its microenvironment, supporting biomarker discovery and translational research.