Early Lung Cancer Detection Using Nucleotide Transition Probabilities in Plasma Cell-Free DNA.
Summary
Lung cancer is a lethal malignancy urgently requiring effective early detection strategies, as current cfDNA-based approaches often lack sensitivity in early stages. This study developed a novel computational feature, First-Order Transition Probability (FOTP), to capture nucleotide sequential dependencies within cfDNA fragments. Using low-pass whole-genome sequencing data, an SVM model trained with FOTP achieved 73.9% sensitivity for stage I and 81.8% for stage II lung cancer at 95% specificity. This method, which significantly outperforms existing fragmentomic features, is biologically interpretable and offers a scalable strategy for early cancer screening.
Analysis
Clinique: This study presents a significant step forward for early lung cancer detection, a disease often diagnosed at advanced stages. The non-invasive method, based on plasma cfDNA, achieved a sensitivity of 73.9% for stage I and 81.8% for stage II with 95% specificity, which is critical for improving patient outcomes. While this is a validation study on a large cohort (1,036 participants), it would justify prospective phase III clinical trials to confirm its utility in mass screening. Clinical impact could be seen in 3-5 years if further validations are successful. Biomol: The key discovery is the identification of the first 10 base pairs at the 5'-end of cfDNA fragments as containing the most discriminative information, reflecting nuclease cleavage biases and chromatin features. The FOTP feature captures these nucleotide dependencies, providing a biologically interpretable basis. The technology employed is low-pass whole-genome sequencing (WGS) on plasma cfDNA, a non-invasive sample type. This approach could lead to the development of clinical tests for early screening and potentially tissue-of-origin prediction. Bioinfo: The innovation lies in the FOTP computational feature, which models nucleotide transition probabilities, a novel approach compared to existing fragmentomic methods. An SVM model was trained and validated on a cohort of 1,036 participants, demonstrating superior performance (AUC of 0.942) compared to current fragmentomic features. The study highlights the robustness and generalizability of the method across independent cohorts and multicancer validation sets, with potential for tissue-of-origin prediction. If deployed clinically, this algorithm could automate and enhance early cancer screening from liquid biopsies.