A Panel of Circulating Exosomal sncRNAs Associated With Lung Cancer Risk up to 10 Years in Advance

Current lung cancer screening fails to identify many individuals at risk beyond heavy smokers, highlighting the need for additional predictive biomarkers. This study profiles circulating exosomal small non-coding RNAs (sncRNAs) in a prospective cohort of 202 smokers, including 68 incident lung cancer cases with up to 16 years of follow-up. A structured 5 × 5 nested cross-validation framework incorporating feature selection, classifier comparison, and outer-loop evaluation identifies a panel of 31 sncRNA risk biomarkers. The optimized random forest model achieves an area under the curve (AUC) of 0.97 in outer-fold testing (sensitivity = 0.93; specificity = 1.00). The derived risk score remains independently associated with incident lung cancer after adjustment for demographic and smoking variables (odds ratio [OR] = 13.19, 95% confidence interval [CI] 6.83-25.45) and is associated with shorter time-to-diagnosis in competing risks analysis (subdistribution hazard ratio [sHR] = 4.78, 95% CI 3.57-6.41). Associations are observed up to 10 years prior to diagnosis. Target gene analysis implicates pathways relevant to lung tumorigenesis, including PI3K-Akt, p53, and MAPK signaling. These findings suggest that plasma exosomal sncRNAs may contribute to early lung cancer risk assessment and warrant further evaluation in independent pre-diagnostic cohorts.