Lipidomic signatures as predictive biomarkers for early-onset lung cancer: Identification and development of a risk prediction model.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
INTRODUCTION: Lung cancer is the leading cause of cancer-related mortality worldwide. While traditionally associated with older adults, early-onset lung cancer (EOLC) is rising, particularly in Asia, which accounts for 75.9% of global cases. Existing lung cancer screening guidelines primarily focus on older populations, which may result in missed opportunities for early detection in younger individuals. Given its distinct clinical characteristics, EOLC warrants dedicated research and targeted interventions. OBJECTIVES: This study aims to characterize the lipidomic profiles specific to EOLC patients (aged 18-49 years) and develop a biomarker-based predictive model to improve risk assessment and early detection. METHODS: The discovery and validation sets included 117 EOLC cases and 121 non-EOLC controls, all aged 18-49 years. Targeted lipidomics analysis, combined with logistic regression, was performed on plasma samples to identify differentially expressed lipids species. Clustering and pathway analyses were conducted to uncover and visualize the internal signatures of the identified lipids. Key lipids were refined using the LASSO-bootstrap regression method combined with the Boruta algorithm. A random forest model was subsequently employed to develop a robust prediction model for EOLC. RESULTS: A total of 843 lipids were identified, with 60 differentially expressed lipids detected, of which 33 were validated in the validation set. Cluster analysis revealed that passive smoking (OR: 2.75, 95% CI: 1.08-7.29) and current smoking (OR: 15.65, 95% CI: 2.55-142.10) were associated with elevated lipid metabolite profiles in EOLC patients. The validated lipids were further refined using LASSO and Boruta methods, which ultimately selected 6 lipids for inclusion in a prediction model constructed with random forest. This model achieved an area under the curve (AUC) of 0.874 in the validation set. CONCLUSION: Our study identified lipidomic signatures associated with the risk of EOLC, offering potential translational implications for lung cancer prevention strategies.