Background: Bladder cancer(BLCA) has a high recurrence rate and metastasis, and its process is closely related to basement membrane remodeling. we developed an interpretable prognostic model based on metastasis-basement membrane-related genes (MBRGs) to enhance clinical and personalized treatment strategies.
Method: Differentially expressed MBRGs from TCGA and GEO cohorts were analyzed. Prognostic genes were identified by univariate Cox and LASSO regression. A six-MBRG risk model was built and externally validated. SHAP analysis quantified feature contributions. Functional enrichment analyzed via GSEA and KEGG. Immune cell profiles estimated with CIBERSORT and ssGSEA. Immunotherapy response predicted using TMB, TIDE, and mutation frequency. Single-cell and spatial transcriptomics localized key genes to cancer-associated fibroblasts(CAFs).
Results: Through analysis of metastasis and basement membrane-associated DEGs, 18 candidate MBRGs were identified and refined via univariate Cox and SHAP to a 6-gene signature (SERPINF1, DDR2, SLIT2, HSPG2, ECM1, RECK). This signature demonstrated prognostic power with AUCs of 0.638-0.674 in TCGA and 0.602-0.742 in GEO cohorts. A clinical nomogram achieved an AUC of 0.827. The high-risk group exhibited elevated M2 macrophages and TIDE scores(a computational metric for predicting tumor immune evasion and immunotherapy response), whereas the low-risk group showed enriched CD8⁺ T cells. Drug assays indicated dasatinib sensitivity in low-risk patients, and LGK974, LY2109761, and Wnt-C59 in high-risk patients. Single-cell RNA-seq and IHC confirmed CAF-specific overexpression of DDR2 and SERPINF1.
Conclusion: The MBRG-based model effectively predicts BLCA prognosis, integrates mechanisms of basement membrane remodeling, EMT, and immune suppression, and identifies DDR2 and SERPINF1 in CAFs as potential targets for personalized therapy.
© 2026. The Author(s).