Molecular subtyping and prognostic modeling of colon adenocarcinoma based on programmed cell death features: a...

Programmed cell death (PCD) plays a complex and critical role in the progression of colon adenocarcinoma (COAD). Elucidating PCD-related characteristics is expected to provide new insights for tumor subtyping, prognosis assessment, and personalized therapy.

This study integrated multi-omics data from COAD and employed 10 clustering algorithms for molecular subtyping. Based on PCD-related genes, a Programmed cell death signature (PCDS) predictive model was constructed using 113 machine learning algorithms. The focus then shifted to the core gene of the model, TERT. Single-cell and spatial transcriptomic data were incorporated to decipher its cellular localization and regulatory pathways. Finally, the function of TERT was validated through in vitro experiments.

We categorized COAD into 5 molecular subtypes with distinct prognostic differences. Subsequently, we successfully developed an 18-gene PCDS. This model effectively predicted patient risk and overall survival in both the training set and multiple independent validation cohorts. The PCDS was closely associated with the tumor microenvironment, mutation burden, and response to immunotherapy. Single-cell and spatial transcriptomic analyses revealed that the core gene, TERT, was specifically highly expressed in malignant epithelial cells. In vitro experiments confirmed that knocking down TERT significantly inhibited the proliferation, migration, invasion, and clonogenic formation abilities of COAD cells. Mechanistically, TERT may inhibit apoptosis, regulate the cell cycle, and promote proliferation potentially through the E2F, G2/M checkpoint, and MYC signaling pathways.

This study defines novel molecular subtypes of COAD through multi-omics clustering analysis and develops a robust PCD-related prognostic signature. Furthermore, it reveals the significant value of TERT as a potential therapeutic target in COAD.