A multi-omics analysis integrating mendelian randomization, brain functional connectivity, and transcriptomics to explore risk-associated features in non-small cell lung cancer.

[BACKGROUND] NSCLC remains one of the leading causes of death from cancer worldwide. Although brain metastases are indeed common and also a significant issue in some cases, what actually causes different patients with NSCLC to develop this condition or not has yet to be clarified by us so as to have a better understanding of the risk.

[METHODS] By integrating multi-omics data, we investigated biological factors associated with the risk of non-small cell lung cancer (NSCLC). Using genetic instruments derived from large-scale GWAS data, Mendelian randomization was conducted to evaluate associations between intrinsic brain functional connectivity and disease susceptibility. Tumour transcriptomic profiles were analyzed using differential expression analysis and weighted gene co-expression network methods to identify disease-associated gene modules. The candidate feature genes were selected by combining the least absolute shrinkage and selection operator (LASSO) regression algorithm with the random forest algorithm. Using the aforementioned genes to construct a predictive model for verification is feasible. It was determined based on the performance of receiver operating characteristics analysis, decision curve analysis and calibration plots.

[RESULTS] Mendelian randomization analysis shows that there are changes in the intrinsic brain connection. Especially in the precuneus and the angular gyrus and cingulate areas. Based on the above, it can be determined that these connection patterns are associated with a risk of developing non-small cell lung cancer. Transcriptomic analysis revealed that there were 204 differentially expressed genes. Among these genes, FRMD3, C7orf68, RCN3, and RBP1 were selected as core features by machine-learning analysis. The ultimate prediction model had good discriminatory capacity for the internal validation; its AUC was 0.873. Analyses based on decision curves and calibration plots can be used to determine if it is appropriate for clinical application.

[CONCLUSION] The results support a link between intrinsic brain functional connectivity and NSCLC risk. Several molecular genes were also identified with predictive relevance. The proposed multi-omics framework offers an exploratory direction for NSCLC risk assessment, though future studies involving independent cohorts are necessary to confirm these findings.