DDR2-COL11A1 Transcriptional Coupling as a Candidate Therapeutic Target in Colorectal Cancer: Integrative Transcriptomic and Deep Learning Validation

Extracellular matrix (ECM) remodeling is a hallmark of colorectal cancer progression, yet the transcriptional mechanisms coordinating collagen deposition and matrix metalloproteinase activation remain incompletely understood. We performed integrated computational analysis of 680 samples across normal mucosa, adenoma, and carcinoma stages to characterize discoidin domain receptor (DDR)-mediated transcriptional networks during tumorigenesis. Stage-stratified correlation analysis of fourteen pathway genes revealed profound divergence between DDR1 and DDR2; DDR1 correlations remained weak across all stages, while DDR2 correlations strengthened 2.59-fold from normal to carcinoma. DDR2-COL11A1 exhibited the most dramatic coupling intensification, increasing from R2=0.007 in normal tissue to R2=0.549 in carcinoma, accompanied by 1.99-fold COL11A1 upregulation. Remarkably, pathway activation occurred despite stable DDR2 expression, indicating enhanced transcriptional coupling efficiency rather than receptor upregulation as the primary mechanism. Deep neural network classification achieved 93.14% accuracy distinguishing disease stages, with SHAP analysis independently validating DDR2-COL11A1 as the most important gene interaction for cancer classification. These findings establish DDR2-specific transcriptional coupling as a functionally important mechanism in colorectal cancer progression and identify COL11A1 as a critical downstream target, suggesting novel therapeutic strategies targeting coupling efficiency rather than receptor abundance.