Methylation of <i>L1RE1</i>, <i>RARB</i>, and <i>RASSF1</i> function as possible biomarkers for the differential diagnosis of lung cancer

Background Lung cancer is the major cause of cancer-related deaths worldwide. Differential diagnosis can be difficult, especially when only small samples are available. Epigenetic changes are frequently tissue-specific events in carcinogenesis and hence may serve as diagnostic biomarkers. Material and methods 138 representative formalin-fixed, paraffin-embedded (FFPE) tissues (116 lung cancer cases and 22 benign controls) were used for targeted DNA methylation analysis via pyrosequencing of ten literature-derived methylation markers (APC, CDH1, CDKN2A, EFEMP1, FHIT, L1RE1, MGMT, PTEN, RARB, and RASSF1). Methylation levels were analyzed with the Classification and Regression Tree Algorithm (CART), Conditional Interference Trees (ctree) and ROC. Validation was performed with additional 27 lung cancer cases and 38 benign controls. TCGA data for 282 lung cancer cases was included in the analysis. Results CART and ctree analysis identified the combination of L1RE1 and RARB as well as L1RE1 and RASSF1 as independent methylation markers with high discriminative power between tumor and benign tissue (for each combination, 91% specificity and 100% sensitivity). L1RE1 methylation associated significantly with tumor type and grade (p<0.001) with highest methylation in the control group. The opposite was found for RARB (p<0.001). RASSF1 methylation increased with tumor type and grade (p<0.001) with strongest methylation in neuroendocrine tumors (NET). Conclusion Hypomethylation of L1RE1 is frequent in tumors compared to benign controls and associates with higher grade, whereas increasing methylation of RARB is an independent marker for tumors and higher grade. RASSF1 hypermethylation was frequent in tumors and most prominent in NET making it an auxiliary marker for separation of NSCLC and NET. L1RE1 in combination with either RARB or RASSF1 could function as biomarkers for separating lung cancer and non-cancerous tissue and could be useful for samples of limited size such as biopsies.