Abstract 1194P
Background
Lung cancer is among the deadliest cancers. Although low-dose computed tomography is recommended for lung cancer screening, it’s accuracy and overdiagnosis is still being debated. Non-invasive cancer screening via circulating tumour DNA (ctDNA) methylation provides an alternative to detect molecular changes in lung cancer cells circulating in the blood. We aimed to identify ctDNA methylation markers of malignant lung lesions to stratify lung nodules, and to develop an early cancer detection model for blood-based non-invasive screening for lung cancer.
Methods
Marker screening was done using a panel of 2200 cancer markers on malignant (N=32) and para-tumour tissues (N=30). Markers with similar distribution of methylation levels among normal tissue, benign nodules, early-malignant nodules, and lung tumours (chi-square test, P>0.001) were discarded. We selected markers with the highest discrimination power by multivariate logistic regression (P<0.001). They were filtered by comparing their median methylation levels in tissues and normal plasma. Filtered markers were tested by classifying lung adenocarcinoma (ADC, N=49) and normal lung tissues (N=8), and plasma from patients with benign nodules (N=39) and ADC (N=35). They were then examined in plasma from benign nodules (N=100), malignant lesions (stage I, N=70; -II, N=12; -III, N=10; and -IV, N=8), or from healthy controls (N=100).
Results
We selected 300 discriminatory markers between malignant and para-tumour tissues, which separate normal tissues, benign and malignant nodules with high accuracy. These markers could classify normal and ADC tissues at 100% accuracy in validation tests and were shown to be present in plasma from lung cancer patients. In further tests, we used lung cancer plasma, plasma of benign nodules caused by inflammation, pneumonia, benign tumours, etc., and normal plasma, and randomly split them into a training (2/3 of samples) and a validation set (1/3) to establish a classification model.
Conclusions
We identified and validated DNA methylation markers accurately classifying benign lung nodules, early cancer, and lung malignancies in lung tissues. They have been tested in plasma from patients with lung nodules and from normal controls to establish a prototype of a non-invasive assay for early lung cancer detection.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
West China Hospital, Sichuan University.
Funding
Singlera Genomics.
Disclosure
Q. He: Shareholder/Stockholder/Stock options: Singlera Genomics. R. Liu: Leadership role, Shareholder/Stockholder/Stock options: Singlera Genomics. All other authors have declared no conflicts of interest.