We conducted a comprehensive proteomic analysis on 115 PDA patient samples, with matched adjacent normal tissue and comprehensive clinical and molecular data. Differential abundance and pathway analyses identified upregulated proteins and pathways within key subgroups. Unsupervised consensus clustering was used to establish a proteomic-based classification of PDA subtypes. A protein-based risk score was developed and validated using a series of multicox regression analyses with LASSO regularization.