Non-apoptotic regulated cell death based prognostic risk model for colorectal cancer using machine learning guided...

Area of Science:

• Oncology
• Computational Biology
• Genetics

Background:

• Colorectal cancer (CRC) prognosis is challenged by tumor heterogeneity.
• Non-apoptotic regulated cell death (NARCD) pathways are emerging as critical factors in cancer progression.

Purpose of the Study:

• To develop a robust prognostic risk model for CRC using a machine learning framework.
• To identify key NARCD pathways and gene signatures for improved CRC patient stratification.

Main Methods:

• A two-step machine learning pipeline was applied to two large, independent CRC patient cohorts (TCGA and E-MTAB-12862).
• 13 NARCD pathways were analyzed using 46 combination ML survival models.
• Logistic regression integrated pathway-specific models to create the combined-regulated cell death index (c-RCDI).

Main Results:

• The synergistic combination of ferroptosis, NETosis, pyroptosis, and autosis showed the highest predictive power.
• The 43-gene c-RCDI model independently stratified CRC patients with high accuracy (5-year AUROC = 0.88, HR = 55.1, P < 0.001).
• Prognostic power was specific to CRC, with distinct biological enrichments and therapeutic sensitivities identified between high- and low-risk groups.

Conclusions:

• The novel two-step ML framework effectively leverages NARCD pathways for CRC prognosis.
• The c-RCDI demonstrates robust predictive ability, offering potential for improved CRC diagnosis and therapy.
• The findings highlight the prognostic significance of NARCD pathways in colorectal cancer.