A mechanosensitive six-gene signature predicts liver metastasis risk in colorectal cancer by integrating mechanobiology with genomics.

The signature is clinically relevant, with elevated SDC1 in tumor tissues compared with normal ones and it remains predictive independently of other clinical variables.

In both training and validation cohorts, the six-gene signature showed strong prognostic performance, with hazard ratios indicating higher risk and time-dependent AUCs of 0.794, 0.775, and 0.815 at 1, 3, and 5 years in the training set.

High-risk tumors are enriched for extracellular matrix organization, collagen-related pathways, integrin binding, and focal adhesion, pointing to ECM/cytoskeleton involvement in metastasis.

Among predictive models, a six-gene StepCox+GBM ensemble delivered comparable performance to more complex models, favoring clinical simplicity.

Experimental validation shows SDC1 is upregulated in colorectal cancer and increases with substrate stiffness; knocking down SDC1 reduces migration and invasion, while stiffness signaling links to focal adhesion remodeling involving paxillin and vinculin.

An intersection of mechanosensitivity and metastasis yielded a 178-gene set, narrowed by univariate Cox to the six-gene signature (SNX7, SDC1, CDCP1, PTGES, EMP2, CALB2) with prognostic value.

Researchers identified 323 mechanosensitive genes responsive to substrate stiffness and connected them with liver metastasis–associated genes to derive 1,708 metastasis-related genes across bulk and single-cell data.

A nomogram combining the six-gene risk score with clinical factors achieved high predictive accuracy (AUC up to 0.848), with solid calibration and demonstrated clinical utility via decision curve analysis.