Abstract

A series of 200 consecutive thin sections were generated from a high-grade serous ovarian tumor and laser microdissected four spatially separated “core” regions of tumor epithelium, along with tumor epithelium, stroma or whole tissue harvests at 200 μm intervals. These distinct tissue collections were analyzed by quantitative proteomics and RNA-Seq. Unsupervised analyses revealed co-clustering of tumor cores with enriched tumor epithelium, which were distinct from the enriched stroma and whole tumor collections. Strong correlations in protein and transcript abundance in tumor epithelium and stromal collections from neighboring thin sections were decreased in samples harvested just hundreds of microns apart. Stroma (mesenchymal) and tumor epithelium (differentiated) displayed a distinct association with ovarian cancer prognostic molecular sub-types with a 2-year difference in median survival. These data reveal substantial tumor microenvironment protein and transcript expression heterogeneity that directly bear on prognostic signatures and underscore the need to enrich cellular subpopulations for expression profiling.