Machine learning assists regulated cell death crucial biomarker selection in adenocarcinoma of the lung: biological data testing and cell assay determination

Han Ning; Ying Jiang; Mengli Zheng; Gao Yang; Lianjun Ma; Yachao Zhao

doi:10.1007/s12672-025-02793-9

Machine learning assists regulated cell death crucial biomarker selection in adenocarcinoma of the lung: biological data testing and cell assay determination

Discov Oncol. 2025 Jul 8;16(1):1285. doi: 10.1007/s12672-025-02793-9.

Authors

Han Ning¹, Ying Jiang², Mengli Zheng³, Gao Yang¹, Lianjun Ma³, Yachao Zhao⁴

Affiliations

¹ Weizikeng Outpatient Department, Jingzhong Medical District, Chinese PLA General Hospital, Beijing, China.
² Department of Clinical Laboratory, The Eighth Medical Center of Chinese, PLA General Hospital, Beijing, China.
³ Department of Thoracic Surgery, The Eighth Medical Center of Chinese, PLA General Hospital, Beijing, 100091, China.
⁴ Department of Thoracic Surgery, The Eighth Medical Center of Chinese, PLA General Hospital, Beijing, 100091, China. zhaoycd999@163.com.

Abstract

Background: Lung cancer is a highly aggressive and lethal cancer requiring prognostic and predictive biomarkers for improving patient outcomes. Here, a prognostic signature for lung cancer was developed and prognostic programmed cell death (PCD)-related genes were identified.

Methods: In this study, we performed comprehensive bioinformatic analyses on diverse datasets, such as Gene Expression Omnibus and The Cancer Genome Atlas. Consensus clustering, Weighted Gene Co-expression Network Analysis, and Gene Set Enrichment Analysis were applied to gain valuable insights from the data, and survival analysis was performed to determine the genes associated with prognosis PCD and construct a prognostic signature.

Results: PCD-related genes (n = 46), significantly correlated with lung cancer prognosis, including ACSL3 and BID, were evaluated. A prognostic gene signature was constructed using 12 genes, which showed excellent overall survival prediction for 1, 3, and 5 years (AUC: 0.687, 0.699, and 0.627). The analysis focused on the nine ley mutant PCD risk model genes and their pan-cancer and elevated mutation frequencies were noted in ALK across several cancer types. The drug sensitivity and immune cell infiltration of the two risk groups were analyzed and revealed noteworthy differences. Patients classified as high-risk demonstrated increased susceptibility to drugs and elevated infiltration of Th2, Tcm, and T helper cells. A prognostic nomogram was developed to predict patient survival at 1, 3, and 5 years, and variables such as age, grading, stage, and the PCD risk score were incorporated. The relationship between PCD-associated genes, genes involved in cell proliferation, and immune cell phenotypes were evaluated. HSPA4 exhibited a significant positive correlation with T helper cells, Th2 cells, and Tcm cells and a negative association with pDCs, TFH, and B cells. In stage III tumors, compared to stage I/II tumors, HSPA4 expression was also significantly upregulated.

Conclusion: Prognostic PCD-related genes for lung cancer were identified and a prognostic signature was constructed. Our findings are invaluable for lung cancer prognostic prediction and personalized treatment.