Aberrant Expression of Collagen Type X in Solid Tumor Stroma is Associated with EMT, Immunosuppressive and Pro-metastatic Pathways, Bone Marrow Stromal Cell Signatures, and Poor Survival Prognosis

Background: Collagen type X (ColXα1, encoded by ) is expressed specifically in the cartilage-to-bone transition, in bone marrow cells, and in osteoarthritic (OA) cartilage. We have previously shown that ColXα1 is expressed in breast tumor stroma, correlates with tumor-infiltrating lymphocytes, and predicts poor adjuvant therapy outcomes in ER/HER2 breast cancer. However, the underlying molecular mechanisms for these effects are unknown. In this study, we performed bioinformatic analysis of -associated gene modules in breast and pancreatic cancer as well as in cells from bone marrow and OA cartilage. These findings provide important insights into the mechanisms of transcriptional and extracellular matrix changes which impact the local stromal microenvironment and tumor progression.

Methods: Immunohistochemistry was performed to examine collagen type X expression in solid tumors. WGCNA was used to generate -associated gene networks in breast and pancreatic tumor cohorts using RNA-Seq data from The Cancer Genome Atlas. Computational analysis was employed to assess the impact of these gene networks on development and progression of cancer and OA. Data processing and statistical analysis was performed using R and various publicly-available computational tools.

Results: Expression of and its associated gene networks highlights inflammatory and immunosuppressive microenvironments, which identify aggressive breast and pancreatic tumors and contribute to metastatic potential in a sex-dependent manner. Both cancer types are enriched in stroma, and implicates bone marrow-derived fibroblasts as drivers of the epithelial-to-mesenchymal transition (EMT) in these tumors. Heightened expression of and its associated gene networks is correlated with poorer patient outcomes in both breast and pancreatic cancer. Common transcriptional changes and chondrogenic activity are shared between cancer and OA cartilage, suggesting that similar microenvironmental alterations may underlie both diseases.

Conclusions: -associated gene networks may hold substantial value as regulators and biomarkers of aggressive tumor phenotypes with implications for therapy development and clinical outcomes. Identification of tumors which exhibit high expression of and its associated genes may reveal the presence of bone marrow-derived stromal microenvironments with heightened EMT capacity and metastatic potential. Our analysis may enable more effective risk assessment and more precise treatment of patients with breast and pancreatic cancer.