Cancer-Associated Fibroblasts Influence Survival in Pleural Mesothelioma: Digital Gene Expression Analysis and Supervised Machine Learning Model

Abstract

The exact mechanism of desmoplastic stromal reaction (DSR) formation is still unclear. The interaction between cancer cells and cancer-associated fibroblasts (CAFs) has an important role in tumor progression, while stromal changes are a poor prognostic factor in pleural mesothelioma (PM). We aimed to assess the impact of CAFs paracrine signaling within the tumor microenvironment and the DSR presence on survival, in a cohort of 77 PM patients. DSR formation was evaluated morphologically and by immunohistochemistry for Fibroblast activation protein alpha (FAP). Digital gene expression was analyzed using a custom-designed CodeSet (NanoString). Decision-tree-based analysis using the “conditional inference tree” (CIT) machine learning algorithm was performed on the obtained results. A significant association between FAP gene expression levels and the appearance of DSR was found (p = 0.025). DSR-high samples demonstrated a statistically significant prolonged median survival time. The elevated expression of MYT1, KDR, PIK3R1, PIK3R4, and SOS1 was associated with shortened OS, whereas the upregulation of VEGFC, FAP, and CDK4 was associated with prolonged OS. CIT revealed a three-tier system based on FAP, NF1, and RPTOR expressions. We could outline the prognostic value of CAFs-induced PI3K signaling pathway activation together with FAP-dependent CDK4 mediated cell cycle progression in PM, where prognostic and predictive biomarkers are urgently needed to introduce new therapeutic strategies.

Keywords

• pleural mesothelioma
• digital gene expression
• cancer-associated fibroblasts
• survival
• machine learning