Early Detection of Non-Small Cell Lung Cancer with Novel 5-Hydroxymethylcytosine DNA Markers: Discovery, Tissue Validation, and Pilot Testing in Plasma

Background: Low-dose computed tomography screening can increase the detection for non-small-cell lung cancer (NSCLC). To improve the diagnostic accuracy of early-stage NSCLC detection, we used ultrasensitive methods to detect cell-free DNA (cfDNA) 5-hydroxymethylcytosine (5hmC) in plasma. Methods: After establishing a model by unbiased whole cfDNA 5hmC sequencing in a training cohort (96 NSCLC (T1N0M0)/95 healthy), the candidate 5hmC model was validated in archival case-controlled internal and external validation cohorts from two independent clinical centers in China. Findings: cfDNA 5hmC patterns were detected in plasma (up 3,823 and down 3,439) more frequently in cases diagnosed with NSCLC than healthy controls (P < 0.001), the functions of which were mainly in the N-Glycan biosynthesis pathway, signaling pathways regulating the pluripotency of stem cells, and the viral carcinogenesis pathway. By applying elastic net regularization to a logistic linear regression model, 100 differential 5hmC loci in gene bodies were selected for cancer classification. Using a weighted diagnosis score computed by the corresponding 5hmC value of the marker genes, their sensitivity and specificity for NSCLC diagnosis were 83-92% and 88-92% in plasma, respectively. The area under the receiver operating curve (AUC) for this panel ranged between 0.968-0.881 in plasma. These results are consistent and comparable with those predicted by carcinoembryonic antigen (CEA) levels (AUC = 0.559-0.570, P < 0.001). Interpretation: We distinguished between the signatures of cfDNA 5hmC in the plasma of early-stage NSCLC patients and healthy controls. We constructed a diagnostic prediction model that showed high diagnostic specificity and sensitivity. Funding: None. Declaration of Interest: All authors have no conflicts of interest to declare. Author Contributions CC and SZ was the principal investigator. YS and XL were responsible for the initial concept of this study. YS, MY, and XL designed the study protocol and oversaw all daily activities of the study. CG, YS, HS, DX, WX and CZ were all responsible for the reporting of individual patient data. YR, and KW did the primary statistical analyses. YR, KW, CG and YS led the interpretation of the data and writing of the manuscript. All authors additionally assisted in data interpretation and preparation of the final article. Ethical Approval: Approval from the institutional review board was obtained prior to initiating the study (k16-264), and all patients and health control people signed an informed consent form.