Digitalized multiparametric analyses of tumor stroma for identification of low perivascular PDGFBR expression and low vessel density as independent prognosis markers for stage IV CRC.

• DOI: 10.1200/jco.2014.32.15_suppl.e14525
• Publication Date: 2019-01-03T17:30:00Z
• Authors: Artur Mezheyeuski; Maja Bradic Lindh; Bengt Glimelius; Per Pfeiffer; Elin Kure; Tone Ikdahl; Eva Skovlund; Halfdan Sorbye; Kjell Magne Tveit; Kristian Pietras; Helgi Birgisson; Jan Mulder; Tormod Kyrre Guren; Anna Portyanko; Arne Ostman
• Keywords: Desmin; Tissue microarray

e14525 Background: Tumor progression and metastasis are regulated by the tumor stroma (TS), suggesting a largely unexplored prognostic significance of its features. The aim this study was to perform multi-parametric analysis colorectal cancer (CRC) explore digitally quantified TS characteristics Methods: A tissue microarray with 247 primary stage IV CRC tumors from phase III NORDIC-VII used. Tumors samples were subjected four double-stainings (CD34 either ASMA, PDGFbR, Desmin or Ki67). Digital images collected image analyses yielding values for 18 different Results: For each case, obtained metrics, related marker-defined subsets fibroblasts pericytes perivascular cells (PC), vascular such as vessel size, total size-categorized density (VD) endothelial cell proliferation We identified strong positive correlations between PDGFbR expression in PC fibroblast-rich compatible common cell-of-origin these cells. coverage not associated size implying independent control aspects angiogenesis. status ASMA- PDGFbR-status that desmin marks distinct set PC. Concerning relevance, two strongest associations observed low (median OS 19.7 mo (N=81) vs. 28.8 high PV (N=81); HR=1.8; 95% CI 1.2-2.7; log-rank p=0.008) VD 18.3 (N=123) 27.6 (N=124); HR=1.7; 1.2-2.3; p=0.001). Both markers remained prognostically significant multivariate including alkaline phosphatase level, WHO performance BRAF mutation status: 95%CI 1.1-2.8; p=0.011 1.2-2.8; p=0.009 respectively. Conclusions: generic procedure multi-parametricdigitalized characterization have been generated used uncover