Evaluation of pyroptosis-associated genes in endometrial cancer utilizing a 101-combination machine learning framework and multi-omics data.

BACKGROUND

Endometrial cancer (EC) is a common and increasingly prevalent gynecological malignancy. Pyroptosis, a pro-inflammatory form of programmed cell death, plays dual roles in cancer but remains poorly understood in the context of EC and its immune microenvironment.

METHODS

We identified pyroptosis-associated genes (PAGs) and applied a 101-combination machine learning framework to construct and validate a robust prognostic model using TCGA bulk RNA-seq and single-cell transcriptomic data. Immune infiltration was assessed using CIBERSORT and Tumor Immune Dysfunction and Exclusion (TIDE), while CellChat was employed to investigate pyroptosis-related cell-cell communication. Drug sensitivity was predicted with OncoPredict.

RESULTS

A seven-gene prognostic model demonstrated robust predictive performance with concordance index (-index) values exceeding 0.70 in both training and validation cohorts. The model stratified EC patients into high- and low-risk groups with distinct immune infiltration profiles and differential responses to programmed cell death protein 1 (PD-1) blockade. Drug sensitivity analysis revealed several therapeutic agents with potential efficacy in high-risk and low-risk subgroups.

CONCLUSION

This study highlights the clinical and immunological relevance of pyroptosis in EC and introduces a PAG-based model with strong predictive and therapeutic potential. These findings provide a foundation for developing pyroptosis-guided precision immunotherapy strategies in EC.