Deciphering lung adenocarcinoma prognosis and immunotherapy response through an AI-driven stemness-related gene signature.

Abstract

Lung adenocarcinoma (LUAD) is a leading cause of cancer-related deaths, and improving prognostic accuracy is vital for personalised treatment approaches, especially in the context of immunotherapy. In this study, we constructed an artificial intelligence (AI)-driven stemness-related gene signature (SRS) that deciphered LUAD prognosis and immunotherapy response. CytoTRACE analysis of single-cell RNA sequencing data identified genes associated with stemness in LUAD epithelial cells. An AI network integrating traditional regression, machine learning, and deep learning algorithms constructed the SRS based on genes associated with stemness. Subsequently, we conducted a comprehensive exploration of the connection between SRS and both intrinsic and extrinsic immune environments using multi-omics data. Experimental validation through siRNA knockdown in LUAD cell lines, followed by assessments of proliferation, migration, and invasion, confirmed the functional role of CKS1B, a top SRS gene. The SRS demonstrated high precision in predicting LUAD prognosis and likelihood of benefiting from immunotherapy. High-risk groups classified by the SRS exhibited decreased immunogenicity and reduced immune cell infiltration, indicating challenges for immunotherapy. Conversely, invitro experiments revealed CKS1B knockdown significantly impaired aggressive cancer phenotypes like proliferation, migration, and invasion of LUAD cells, highlighting its pivotal role. These results underscore a close association between stemness and tumour immunity, offering predictive insights into the immune landscape and immunotherapy responses in LUAD. The newly established SRS holds promise as a valuable tool for selecting LUAD populations likely to benefit from future clinical stratification efforts.