P90. High sensitivity proteomic profiling of archival pancreatic ductal adenocarcinoma defines tumor and tumor microenvironment changes

Introduction

We previously identified two tumor-intrinsic molecular subtypes of pancreatic ductal adenocarcinoma (PDAC) with prognostic and therapeutic significance. We used multiplex inhibitor bead mass spectrometry (MIB/MS) and found that the two molecular subtypes harbor distinct kinomes. However, MIB/MS and related proteomic methods require large quantities of fresh-frozen tissue. New high sensitivity proteomics allows analysis of formalin-fixed, paraffin-embedded (FFPE) tissues. We determined the feasibility of these new technologies for resected PDAC FFPE samples.

Methods

De-identified FFPE PDAC samples were sectioned at 4μm and H&E stained. After pathologist review, paired macrodissected (MD) tumor and whole tumor sections (WTS) underwent Orbitrap Astral mass spectrometry. Proteins detected in at least 2 of 3 replicates were retained and log₂-transformed. Differential expression between regions was evaluated with a moderated t-test (limma) and a ≥2-fold change threshold (1.5-fold for kinases). Kinases and proteins were compared to previously identified PDAC-subtype associated kinases and Moffitt tumor gene sets.

Results

We identified ~7500 proteins and 310 kinases across all samples. >90% of proteins were consistently detected in ≥2 replicates. Tumor regions demonstrated significant (p < 0.05) upregulation of kinases EPHB2, CSNK1G2, and STK10, all previously found to have PDAC subtype-specific protein expression via MIB/MS. Moffitt tumor-intrinsic proteins were similar between MD and WTS. However, MD tumor sections had significantly different global protein expression for 356 proteins compared to WTS (p < 0.05, log2FC > 1). Gene set enrichment analysis showed enrichment of epithelial mesenchymal transition and complement gene sets in the WTS compared to the MD tumors (p < 0.001, log2FC > 0.5). Gene sets upregulated by myc and cellular stress were enriched in the MD tumors vs the WTS (p < 0.001, log2FC > 0.5). Furthermore, SERPINB3 from the Moffitt 50 signature was highly expressed in MD but not WTS (p < 0.001, log2FC > 10), suggesting that some tumor-specific genes may be missed in the absence of MD.

Conclusions

Our results show that FFPE proteomics recapitulate prior tumor and kinase signatures in both WTS and MD samples. Tumor regions display consistent enrichment of molecular subtype and subtype-specific kinases reflective of PDAC tumor biology. MD when feasible may further elucidate tumor and tumor microenvironment changes. This approach supports the feasibility and promise of FFPE proteomics in future studies.