Analysis of Multiple Programmed Cell Death Patterns and Functional Validations of Apoptosis-Associated Genes in Lung Adenocarcinoma.

Background: Lung adenocarcinoma (LUAD) is marked by its considerable aggressiveness and pronounced heterogeneity. Programmed cell death (PCD) plays a pivotal role in the progression of tumors, their aggressive behavior, resistance to treatment, and recurrence of the disease.

Methods: Using expression data from 878 patients across four multicenter cohorts, we identified 13 consensus prognostic genes from 1481 genes associated with PCD. We employed 10 machine-learning algorithms, generating 101 combinations, from which the optimal algorithm was chosen to develop an artificial intelligence-derived cell death index (CDI) on the basis of the average C-index.

Results: The training cohort and three external validation cohorts consistently demonstrated that CDI could accurately predict LUAD prognosis. Moreover, CDI showed significantly greater accuracy than traditional clinical variables, molecular characteristics, and 22 previously published signatures. Patients in the low-CDI group had a more favorable prognosis, higher levels of immune cell infiltration, better responsiveness to immunotherapy, and a higher likelihood of displaying the "hot tumor" phenotype. Single-cell analysis revealed that neutrophils had the highest CDI scores and exhibited significant differences in marker gene expression.

Conclusions: Pseudotime trajectory analysis indicated that BCL2L14 plays a crucial role in the developmental pathway of neutrophils, potentially influencing the fate of LUAD cells. Knockdown of BCL2L14 significantly reduced the growth, proliferation, and colony formation abilities of LUAD cells, while also enhancing apoptosis rates.