Abstract Background: Timely detection plays a pivotal role in lung cancer management. This study was conducted to develop ‘LedX’, a blood-based methodology for the early screening of lung cancer. Methods: Patients with lung cancer or benign lung disease and healthy individuals were enrolled. Quantitative polymerase chain reaction was employed to measure cell-free DNA (cfDNA) methylation levels. The differentially methylated regions were ranked according to the area under the receiver operating characteristic curve (AUC–ROC), and the top 50 markers were selected as candidates. A total of 13 markers were incorporated into the LedX model, three of which were obtained from a literature. LedX was subsequently trained on 325 samples (133 cancer, 63 benign, and 129 normal), then tested using an independent set comprising 317 samples (151 cancer, 134 healthy, and 32 benign). Besides, the performances of LedX plus low dose computed tomography (LDCT) were also tested. Results: The LedX model exhibited an overall sensitivity of 0.820/0.709 (training/test) and specificity of 0.817/0.910 (training/test). The three submodels, LedX-CN (cancer versus normal), LedX-CNB (cancer versus normal + benign), and LedX-CB (cancer versus benign), reached an overall AUC–ROC of 0.897–0.942/0.885–0.933 (training/test). The efficiency was generalized across multiple populations under different health/disease status scenarios. LedX plus low LDCT demonstrated similar specificity to that observed with LDCT or LedX alone. LedX plus LDCT achieved higher sensitivity in diagnosing Stage I cancer compared to LDCT or LedX alone. Conclusions: LedX, a noninvasive circulating tumor DNA methylation assay, demonstrated satisfactory sensitivity and specificity for lung cancer detection; thus, the model can be implemented in various clinical settings. Thus, LedX represents a potentially simple and cost-effective tool for the early screening and diagnosis of lung cancer.
Zhang et al. (Wed,) studied this question.