Advancing lung adenocarcinoma prognosis and immunotherapy prediction with a multi‐omics consensus machine learning approach

• DOI: 10.1111/jcmm.18520
• Publication Date: 2024-07-03T13:40:20Z
• Authors: Haoran Lin; Xiao Zhang; Yanlong Feng; Zetian Gong; Jun Li; Wei Wang; Jun Fan
• Keywords: Omics

Lung adenocarcinoma (LUAD) is a tumour characterized by high heterogeneity. Although there are numerous prognostic and immunotherapeutic options available for LUAD, dearth of precise, individualized treatment plans. We integrated mRNA, lncRNA, microRNA, methylation mutation data from the TCGA database LUAD. Utilizing ten clustering algorithms, we identified stable multi-omics consensus clusters (MOCs). These were then amalgamated with machine learning approaches to develop robust model capable reliably identifying patient prognosis predicting immunotherapy outcomes. Through two prognostically relevant MOCs identified, MOC2 showing more favourable subsequently constructed MOCs-associated (MOCM) based on eight MOCs-specific hub genes. Patients lower MOCM score exhibited better overall survival responses immunotherapy. findings consistent across multiple datasets, compared many previously published LUAD biomarkers, our demonstrated superior predictive performance. Notably, low group was inclined towards 'hot' tumours, higher levels immune cell infiltration. Intriguingly, significant positive correlation between GJB3 (R = 0.77, p < 0.01) discovered. Further experiments confirmed that significantly enhances proliferation, invasion migration, indicating its potential as key target treatment. Our developed accurately predicts patients identifies beneficiaries immunotherapy, offering broad clinical applicability.