Blood-based DNA methylation panel for the early diagnosis of lung cancer.

Abstract

e20551 Background: Early detection of lung cancer (LC) is vital for reducing of LC-related mortality. Although the screening of persons at high risk for LC by low-dose computed tomographic (LDCT) has showed an inspiring 20.0% decrease in mortality, it also comes along with a dramatically high false positive rate for distinguishing the malignant nodules from the benign nodules. Thus, it is necessary to develop a diagnostic tool with high discriminant ability for identifying malignant nodules. Methods: The methylation levels of five candidate genes were quantitatively measured via mass spectrometry. A diagnostic model was developed by blood-based methylation levels in training cohort (650 LC cases vs. 114 benign lung nodules), and further validated the performance in validation cohort (195 LC cases vs. 39 benign lung nodules) by binary logistic regression. Notably, 91.8% LC cases were collected at Stage I. Results: The methylation model of 35 CpG sites was developed and validated for LC diagnosis. This model achieved a sensitivity of 85.4% and a specificity of 92.5% in training cohort [Area Under the Curve (AUC): 0.945], and a sensitivity of 84.1% and a specificity of 97.2% in validation cohort (AUC: 0.932). The performance was well maintained in (a) Stage I subgroup (n = 447), with a sensitivity of 82.6% and a specificity of 81.6%; (b) nodule diameter ≤ 1 cm (n = 113), with a sensitivity of 80.6% and a specificity of 81.6% in validation cohort. Conclusions: This study suggests that the blood-based DNA methylation panel may provide the potential utility for early diagnosis of LC cases, which would promote early diagnosis and benefit more LC patients.