Integrating zinc homeostasis network and immune landscape: a five-gene prognostic framework for precision oncology in...

Lung adenocarcinoma (LUAD) exhibits marked heterogeneity in clinical outcomes and therapeutic responses, underscoring the imperative for reliable prognostic biomarkers. Dysregulation of zinc homeostasis is an emerging hallmark of cancer, contributing to tumor progression through multifaceted mechanisms including exacerbated oxidative stress and sustained oncogenic signaling. This study aimed to develop and validate a novel prognostic signature based on zinc homeostasis network-related genes for stratifying LUAD patients into distinct risk groups to predict clinical outcomes and inform therapeutic strategies.

Transcriptomic and clinical profiles of LUAD cases from the TCGA database were integrated to screen for differentially expressed genes (DEGs) involved in zinc homeostasis network. A prognostic risk model was constructed via univariate, multivariate, and LASSO regression analyses, and externally validated using GEO datasets. Model performance was evaluated using a nomogram, time-dependent ROC curves, and decision curve analysis. To characterize immune microenvironment heterogeneity across risk subgroups, we applied seven deconvolution algorithms, ssGSEA and single-cell profiling. Spearman correlation analysis and Wilcoxon rank-sum tests were used for investigating the associations between risk stratification and immunomodulatory markers, tumor mutational burden (TMB), as well as the predicted responsiveness to conventional chemotherapeutic and targeted therapies. In vitro experiments validated the expression levels of key candidate genes and confirmed their biological functions.

Differential expression analysis identified 124 zinc homeostasis network-related DEGs, of which 16 showed significant prognostic relevance. A five-gene risk model stratified patients into distinct prognostic groups, with high-risk cases showing markedly reduced overall survival. Risk score correlated positively with advanced clinical stages. Multivariate Cox regression confirmed the model as independently predictive of LUAD prognosis. A nomogram integrating risk score and clinical features was constructed for predicting 1-, 3-, and 5-year survival. Immune profiling showed that low-risk cases had hot tumor phenotypes, elevated immune scores and infiltration by immune cells, while those at high risk showed raised levels of immunotherapy resistance markers and increased TMB. Drug sensitivity analysis indicated differential responses to chemotherapeutic and targeted agents across risk groups. Independent knockdown of SLC16A3 or overexpression of EGR2 significantly suppressed malignant behaviors in LUAD cells. Additionally, SLC16A3 downregulation reduced cisplatin sensitivity in LUAD cells.

Our study highlights the clinical translational potential of a five-gene zinc homeostasis network signature in predicting prognosis and guiding personalized therapeutic strategies for LUAD.