Development and validation of an interpretable prognostic model for bladder cancer based on lactylation associated...

Area of Science:

• Oncology
• Epigenetics
• Bioinformatics

Background:

• Bladder cancer (BLCA) has high recurrence and poor outcomes, necessitating novel biomarkers.
• Lactylation, a new epigenetic modification, presents a potential avenue for biomarker discovery.
• Precision medicine approaches require robust prognostic tools for BLCA management.

Purpose of the Study:

• To construct and validate a prognostic model for BLCA utilizing lactylation-related genes.
• To investigate the relationship between the prognostic model and the tumor immune microenvironment (TIME).
• To evaluate the model's association with tumor mutational burden (TMB) and drug sensitivity.

Main Methods:

• Integrated transcriptomic and clinical data from GEO and TCGA-BLCA cohorts.
• Employed univariate Cox, LASSO, and multivariate Cox regression to identify core prognostic genes.
• Utilized SHapley Additive exPlanations (SHAP) for interpretable survival predictions and analyzed TIME, TMB, and drug sensitivity.

Main Results:

• A 4-gene risk model (ATAD3A, MKI67, VWF, CCL2) was established with robust predictive efficacy in training and validation cohorts.
• The risk score independently predicted prognosis, and SHAP analysis provided personalized survival risk interpretability.
• High-risk group showed M2 macrophage enrichment and high CCL2 expression; low TMB and high-risk patients had the poorest outcomes.

Conclusions:

• This is the first study to establish a 4-gene lactylation-related prognostic model for BLCA using a LASSO-SHAP framework.
• The model offers precise, visualizable predictions for individualized survival risk in BLCA.
• Findings elucidate the lactylation-driven immunosuppressive microenvironment, supporting personalized immunotherapeutic and targeted strategies.