Integrated analysis identifies a palmitoylation-associated prognostic model (ACSM5/SKA3) for lung adenocarcinoma across multiple cohorts

Background Lung adenocarcinoma (LUAD) is the most common subtype of non-small cell lung cancer and remains associated with poor prognosis despite therapeutic advances. Protein palmitoylation, particularly reversible S-palmitoylation, modulates oncogenic signaling, membrane localization, and immune checkpoint stability, but the prognostic landscape of palmitoylation-related genes (PRGs) in LUAD and their relationship with the tumor microenvironment are not well defined. Methods PRGs were curated from GeneCards and integrated with transcriptomic and clinical data from The Cancer Genome Atlas Lung Adenocarcinoma (TCGA-LUAD) cohort. Differential expression and univariate Cox analyses were performed to identify prognostic PRGs, which were then used for consensus clustering to define molecular subtypes and to construct a LASSO-Cox prognostic model. Associations between the risk score and functional pathways, genomic alterations, tumor mutational burden (TMB), tumor immune dysfunction and exclusion (TIDE) scores, and immune infiltration were evaluated. Three Gene Expression Omnibus (GEO) cohorts were used for external validation. Single-cell RNA sequencing (scRNA-seq) datasets and reverse transcription-quantitative polymerase chain reaction (RT-qPCR) of paired LUAD and adjacent tissues were used to assess cell type–associated expression and validate key genes. Results We identified 201 differentially expressed PRGs in LUAD and 15 genes significantly associated with overall survival, including the protective gene ACSM5 and the risk gene SKA3. PRG-based consensus clustering defined two molecular subtypes with distinct prognoses, differential enrichment of metabolism- versus proliferation-related pathways, and divergent immune microenvironment features. A two-gene signature comprising ACSM5 and SKA3 stratified patients into high- and low-risk groups in The Cancer Genome Atlas (TCGA) and three external cohorts, with time-dependent area under the curve (AUCs) of 0.717, 0.733, and 0.697 at 1, 3, and 5 years in TCGA and comparable performance in GEO validation cohorts (5-year AUC up to 0.778), indicating moderate discriminative performance. Low-risk patients generally showed higher stromal and immune scores and increased infiltration of multiple immune cell subsets, along with complex patterns in TMB and TIDE indices. Single-cell analyses indicated predominant expression of ACSM5 in macrophages and of SKA3 in proliferating CD8+ T cells and macrophages, and RT-qPCR confirmed downregulation of ACSM5 and upregulation of SKA3 in LUAD tissues. Conclusion This integrated analysis outlines a palmitoylation-associated molecular framework in LUAD and supports an ACSM5/SKA3 two-gene signature as a candidate prognostic tool that stratifies patient outcomes and reflects distinct metabolic, proliferative, and immune microenvironmental states, potentially providing a palmitoylation-associated basis for risk stratification and for the design of future mechanistic and therapeutic studies in LUAD.